# Falnor: full site text Source: https://falnor.io. Generated 2026-10-28. Plain text of every public page. # Home (https://falnor.io/) ## We make work flow on its own. Invoices wait for a signature. Customers wait for a reply. Your team spends the day chasing. We build the system that keeps that work moving, on the software you already use, and it checks with a person before anything important. ## Work does not fail. It stops. An invoice, a claim, a new customer. The real work takes minutes. Then it sits and waits: for a missing detail, a sign-off, or someone to remember it. Your best people keep it moving by hand. That is not their fault. It is how the work is set up. - 58% of the working day goes on chasing, searching and switching between apps, not on the job itself. Source: Asana, Anatomy of Work, 2023. 9,615 office workers in six countries, including the UK.. - Nearly 4 hrs a week lost just finding your place again after switching between apps. Source: Harvard Business Review, 2022. ## Good work needs a good system, not more effort. Your business already has the tools: email, a customer database, accounts software, spreadsheets. What is missing is what joins them up. Where each detail comes from, which rule applies, who signs off, and what gets updated. Today that lives in the heads of a few people. We write it down with your team, then build it into a system that runs on your own software. We start small. At first the system only watches and suggests, while your team carries on as normal. It takes on work only once it has shown it gets things right, and it always stops for a person on the big decisions. It grows one proven step at a time, and it belongs to you. ## One task, from chased by hand to running by itself. Yesterday a customer refund sat in four places: an email, the customer record, the accounts and a chat waiting for a yes. Someone senior pushed it through each one. Today the system checks all four, applies your refund rule, issues the credit and closes the case. The one unusual refund lands with the right person, with every detail ready. They approve it from their phone, and the system carries on. Their afternoon goes on the work they were hired for. ## Our capabilities ### It looks before it acts At first the system can only read your software, never change it. It earns the right to update anything one task at a time, once it has shown it reads your real cases correctly. ### It stops where a person should Anything important, unusual or hard to undo waits for a named person to say yes. You decide where that line sits. The system waits as long as it takes. ### It shows its working Every step comes with proof: what it looked at, which rule it followed and who approved it. When someone asks what happened, the answer is already written down. ### It belongs to you It runs on your own systems, and your data stays with you, in the UK or EU if that is where you keep it. Your team can see every rule. If we disappeared tomorrow, it would keep running. ## One system. Three things you can see. How much of the work now runs by itself, how it connects to your software, and how it earns more responsibility over time. ### The share of work that runs by itself. This is the number we are judged on: how many tasks go from start to finish with nobody chasing them. We agree the target with you at the start and prove it on your own cases before anything goes live. ### Works with your software It connects to the tools you already use and starts by only reading from them. It is allowed to update them one task at a time, once it has earned your trust. ### Trust, earned step by step Like a new hire, it starts by watching, then suggests, then helps, then runs the routine work. Your team signs off each step up. If it ever gets something wrong, it steps back automatically. ## What the system does with every task. Whether it is an email, a customer record, the accounts or a document, the job always comes down to the same four steps. Today a person does them by hand, between tabs, when they find the time. The system does them the same way every time, and stops for a person when it should. ### Gather the details Pull what it needs from your email, customer records and accounts, and keep it all together, so nobody starts from a blank page. ### Check the rule Compare the amount with your policy and past approvals. Under the limit, it goes ahead. Over the limit, it waits for the right person. ### Finish the job Update the accounts, send the customer a note and file the paperwork. The task is closed, with a clear reason, not left as a draft. ### Ask a person Anything unusual comes to your team with the details ready. Approve it, change it or hold it. The system remembers the decision next time. ## What your team gets back. Less time chasing, more time on the work that needs them. Only real decisions reach a person, and every finished task explains itself. - Hours back every week, because the system does the chasing, searching and copying. Your people spend their time on judgement, not updates. - People only on decisions, because routine tasks finish on their own. Approvals and unusual cases reach your team with the details already attached. - A clear reason on every task, showing what it looked at, the rule it followed and who signed off. Ready when your auditor or a customer asks. ## Off-the-shelf tools give you parts. We hand over a system that works. Most options leave the hard part with you: the rules, the exceptions and who decides. We build around how your business actually works, in your words and to your limits. ### Twice as likely to go live Share of projects that go live, by who built them. MIT, The GenAI Divide, 2025. Interviews with 52 organisations. - Built in-house: 33% - With a specialist partner: 67% ### Only one option covers it all What each kind of tool takes off your team. - Automation tools: Exceptions left with you, Your rules partly, Joined up handled, Goes live handled - DIY platforms: Exceptions left with you, Your rules left with you, Joined up handled, Goes live partly - Single-task apps: Exceptions partly, Your rules partly, Joined up left with you, Goes live handled - Built in-house: Exceptions handled, Your rules handled, Joined up handled, Goes live left with you - Falnor: Exceptions handled, Your rules handled, Joined up handled, Goes live handled ### Right 9 times in 10 first Tested on your real cases before it goes live. ### 70–85% then runs by itself The rest waits for a person, with the details ready. ### Freedom is earned It takes on more only once the last step is proven. ### Built on the tools you have Usually 3 to 6 of them. Your data stays where it is. We agree the goals with you at the start and test them on your cases before go-live. Price is set after we map your work. ## Start with a map, not a commitment. We begin by mapping one part of your business with your team. We look at how the work really moves, where it gets stuck, and what a system would need to run it. You keep the map, whether or not we build anything. - The steps: Every step and hand-over, as the work really happens, not as the manual says. - The rules: Every decision rule, including the ones only one person knows. - The software: Each tool the work passes through, and what the system would need access to. - The hold-ups: Where work waits, for how long, and who it is waiting on. - The starting point: How much work there is, how long it takes and what it costs today, measured before we build anything. The map sets the scope and the price. No obligation to go further. ## Straight answers. The questions finance, IT and procurement teams usually ask, answered before the call. ### Will this replace my team? No. It takes over the chasing, not the decisions. Your people keep the judgement calls, and the system learns from them. ### Is this like Zapier or a bot that clicks through screens? No. Those tools repeat one step you already understand. We build the whole system around the work: the unusual cases, the unwritten rules, the points where a person decides, and a record of everything. Some steps get automated along the way, but the system is what we deliver. ### We tried a no-code or AI tool and it only half works. That is a common place to start. These tools handle the easy cases; the exceptions and the checks are where they stall. We look at what your tool does today and what still lands on people. Then we decide with you whether to extend it, replace it or build around it. ### How is this different from buying an AI platform? A platform gives you building blocks and leaves the hard part, your rules, your limits and your checks, to you. We build the finished system, test it on your real cases and hand it over. ### Our process is not written down anywhere. Is that a problem? Not at all. In most businesses it lives in people’s heads. Mapping writes it down: where each detail comes from, the rule at each step, who decides and what gets updated. You keep that map whether or not we build. ### Why is there no price on the site? Every business works differently, so there is no price list. We map your work first. That sets the scope, and you get the scope, the timeline and the price in writing before anything is built. ### Why start small? Because systems that work grow from small ones that worked. We start with one task, one person who owns it and one goal we can measure. If we cannot agree those on the call, we will say so. ### Who owns what you build? You do. The system, its rules and its full record are handed over and run on your own systems. You can keep using it without us. ### Where does our data live? With you. The system runs on your own systems and reads from them, so your records stay where they are, including in the UK or EU. We do not copy your data into ours. ### Is it GDPR-friendly? It is built to be. Your data stays in your systems and under your control. The system only gets the access each task needs, and every action is recorded. That makes it easier for your data protection team to answer questions. ### Will my team be flooded with approval requests? The opposite. There are more at the start. Each time your team decides the same thing the same way, the system learns it, and the requests go down week by week. ### What if the AI underneath changes? It does not matter much. Your rules and checks sit outside the AI, where your team can read them. We keep a set of your real past cases and re-test the system before anything underneath changes. # Approach (https://falnor.io/approach) ## The AI is the least interesting part. A Falnor system is built to be relied on. It looks before it acts, stops before anything that cannot be undone, and keeps a record of every decision. These seven principles hold true whichever AI sits underneath. ## How one task moves. Every task goes through the same six steps. The first three change nothing. The check in the middle decides whether the last three are allowed to happen. - Arrives: The task comes in with its own reference, so it can never be handled twice by mistake. - Looks: The system gathers every detail it needs. Nothing is changed yet. - Decides: Your rules are applied to what it found. The result says which rule was used. - Checks: If the task is over your limit or unusual, it stops and asks a named person. It waits as long as it takes. - Acts: It updates each system once. If something fails, it tries again. Nothing is left half-done. - Records: What changed, where the details came from, the rule and who approved it are all saved for you to see. ## What every Falnor system is built on. None of these depend on a particular AI. Together they are why a system can be trusted to keep work moving without someone pushing it. ### 01 Look before acting At first, the system can only read your software, never change it. It is allowed to update things one system and one task at a time, once it has shown it understands your real cases. What it means for you: Nothing in your business can be changed by a system that has not first proven it reads things correctly. ### 02 Pause safely before big steps Before anything that cannot be undone, the system saves exactly where it is. If it needs a person, it waits at that point and picks up from there when they reply, even hours later or after a restart. What it means for you: An approval is never lost, and nothing is ever done twice. ### 03 Clear rules decide, not the AI The AI never approves its own actions. Each time it wants to do something, a separate set of rules checks it and answers yes, ask a person, or no. Your team can read those rules. What it means for you: Permission comes from rules you can see, not from an AI that can be talked round. ### 04 Proof, not guesswork A confidence score is only a feeling. Every step is checked against what actually happened: what the system looked at, what it changed and the rule that allowed it. If something does not add up, it is flagged, not hidden. What it means for you: When your auditor or a customer asks why, you can show them the record. ### 05 Trust is earned and can be taken back Every system starts by only watching and suggesting, while a person still does the work. It is given more to do in small steps, only when the results prove it is ready. Your team signs off each step up. If it ever gets something wrong, it steps back automatically. What it means for you: Trust is built on a track record, not switched on in the settings. ### 06 Rules come from your decisions When your team answers the same question the same way, the system suggests turning it into a rule. Someone approves it, and fewer requests need a person, for a reason you can see and undo. What it means for you: The number of approvals your team handles goes down week by week, and every drop is explained. ### 07 It runs where your data lives It runs on your own systems, so your data stays where it is, in the US, UK or EU if that is where you keep it. Its links to your software, its rules, its records and your test cases all belong to you. If the AI underneath changes, we re-test on your cases before anything goes live. What it means for you: You can change the AI, the supplier, or us, and keep the system. ## Why most AI projects stall, and what the successful ones do first. Independent research tells the same story. Businesses are adopting AI faster than they are putting checks around it, and most projects never get past the trial. The problem is rarely the AI itself. It is how the work is mapped, checked and owned, which is what Falnor is built around. - 40%+ of AI projects that act on their own will be scrapped by the end of 2027, mostly on cost or unclear value. Source: Gartner, June 2025. - 5% of company AI trials make a measurable difference to the bottom line. Source: MIT, The GenAI Divide, 2025. ### Businesses are adopting AI faster than they are putting checks around it Share of organisations. Deloitte, State of AI in the Enterprise, 3,235 leaders, 2026. - Expect to use AI that acts on its own within two years: 74% - Use it today: 23% - Have proper checks and rules in place for it: 21% ### Who builds it decides whether it goes live Share of projects that went live. MIT, The GenAI Divide, 2025. - Built with a specialist partner: 67% - Built in-house: 33% - Businesses with weaker checks are more likely to find out about an AI mistake only after it has caused a problem. - Many businesses using AI do not review everything it produces before it is used. - Many products sold as AI assistants are older automation or chatbots with a new label. It is worth asking what is underneath. ## Everything in the system is one of nine things. If you know these nine words, you can follow everything the system does. Your industry supplies the name for the task; the rest stays the same. - Item: The thing being worked on. A claim, a support request, an invoice. - Workflow: The steps an item goes through, in order. - Run: One item going through its workflow once. - Step: One small action inside a run, such as checking a record. - Task: An item and where it has got to, as your team sees it. - Decision: A point where the system stopped and asked a person. - Rule: A standing answer, learned from your team’s past decisions. - Level: How much the system is trusted to do on its own. Earned, not set. - Evidence: What the system looked at and where it found it. Never a guess. ## Built on your systems. Yours from day one. Handing it over is part of the job, not an extra. Everything the system runs on stays with you, and your team knows how to run it. - From day one, it runs on your systems: Your software, your accounts, your data. Nothing moves across to us. - At handover, your team takes the keys: A walkthrough and the exit pack: every connection, rule and test case, written in plain files. - Any day after, it keeps going without us: Your team can change a rule, swap the AI or switch it off in one click. No call to us needed. # About (https://falnor.io/about) ## We build the systems your business runs on. Falnor designs and builds systems that keep everyday business work moving. We map how your work really runs, then build a system around it, on the software you already use. Your people stay in control until the system has proven it can be trusted. ## The work is not glamorous. It is held together by hand. Work passes through emails, documents, websites, spreadsheets and your main business software. At every hand-over, details get lost. It is not one big problem. It is a hundred small hold-ups a week, each one landing on someone experienced enough to sort it out. Together they cost real money and cause real frustration. - Chasing sign-offs by email, chat and online portals - Problems landing on whoever replies first, not whoever should own them - Typing the same details into three different systems - Month-end preparation that starts from scratch every time - A rule everyone follows that nobody ever wrote down - The one detail that would finish the job, sitting in someone else’s inbox None of this needs cleverer AI or another app. It needs a system: your tools joined up, your rules written down, and a clear point where a person decides. That is what we design and build. ## One system, started small and grown on results. We start with one part of your business. We map where work gets stuck today and write down each step from start to finish. Then we build the system on the software your team already uses. It gathers the details, applies your rules, asks the right person when it should, updates your records and keeps proof of every step. It is not a platform you set up yourself, and it is not another trial that goes nowhere. It is a system we design with you, that your team uses every day. We show it does what we agreed, then hand it over so it is fully yours. - On your computer: See what is open, what needs you, and exactly what happened. - On your phone: The one decision that needs you comes to your phone. The system waits, then carries on. - In Slack, Microsoft Teams and email: Approvals and alerts arrive in the tools your team already opens, with the same record behind each one. ## Good work needs a good system. And ours start with no freedom at all. Most businesses already have the tools. What is missing is what joins them up: where each detail comes from, which rule applies, and who decides. Today that lives in the heads of the people who keep things moving. We write it down and build it into a system. At first the system only watches and suggests, while your team carries on as normal. It takes on work one small step at a time, once it has shown it gets things right. If it ever gets something wrong, it steps back automatically. Your experts stop chasing work and start making decisions. ## Work gets stuck far more often than it fails. These figures are not ours. They show the size of the problem: most of the working day goes on pushing work along by hand. - 58% of the working day goes on chasing, searching and switching between apps, not on the job itself. Source: Asana, Anatomy of Work, 9,615 office workers in the UK, France, Germany, US, Australia and Japan, 2023. - Nearly 4 hrs a week lost just finding your place again after switching between apps. Source: Harvard Business Review, 2022. ## Building a system people rely on is a different job from showing off a tool. Behind every project is careful work. At first the system can only look at your software. Your people stay in control, and we test on your real cases before anything goes live. You are left with a clear record your team can rely on without calling us. You will not see most of it. You will see the work moving. - It looks before it acts: At first the system can only read your software. It earns the right to change anything once it has proven itself on your cases. - Tested on your real work: It runs alongside your team on real cases before it goes live, and it is re-tested whenever anything changes. - A person for anything important: Above the limit you set, the system stops and waits for a named person to say yes. - A record your team owns: Every step, change and approval is kept where you control it, and your team can read it without us. ## Thirty minutes. One process. A straight answer. Tell us where work gets stuck. If a system will not fix it, we will tell you on the call. # Showcase (https://falnor.io/showcase) ## See the system at work. Four short looks inside a Falnor system: the morning summary, the decisions that wait for a person, and where every task has got to. Video: Falnor Desk: your morning summary, under a minute (https://falnor.io/videos/falnor-desk.mp4). Video: Falnor Ask: questions answered from live data, under a minute (https://falnor.io/videos/falnor-ask.mp4). Video: Falnor Workflows: an agent on every step, under a minute (https://falnor.io/videos/falnor-workflows.mp4). Video: Falnor Connect: your apps, your channels, under a minute (https://falnor.io/videos/falnor-connect.mp4). Product walkthrough. Names, figures and records are sample data. ## Why it works. It is not another screen to keep an eye on. It keeps the work moving, and only brings a person in when a real decision is needed. ### Your day starts with a summary, not a spreadsheet. Each morning you see what the system did, what needs you, most urgent first, and what is stuck. Nobody has to go chasing for updates. ### Only the decisions come to you. Routine tasks finish on their own. Anything important, unusual or hard to undo waits for the right person, with the details already gathered. Approving takes one click. ### You see where the work is, not just that it is late. Each step shows how many tasks have passed through and which are waiting or stuck. You fix the one hold-up instead of chasing the whole process. ### Problems show up before they cause trouble. Spending is tracked against the limit you set, and access that is about to expire is flagged days ahead. You hear about it while there is still time to act. ### Every step leaves a record. What it looked at, which rule it followed and who approved it. When someone asks what happened to a request, the answer is already written down. ### It runs on your software and stays yours. It is built on the systems you already use, and your data stays where it is. Your team can see every rule, and the system keeps running without us. ## Your business will not look exactly like this. That is why every project starts by mapping how your work really moves. You keep the map, whether or not we build anything. # Brand kit (https://falnor.io/brand) ## Brand kit. Logos, colours, fonts, words and tone of voice for partners, press and joint marketing. Everything here serves one idea: work does not fail, it stops, and we make it flow. Tagline: We make work flow. ## Work does not fail. It stops. We make it flow. Falnor designs and builds systems that keep everyday business work moving. We map how a client’s work really runs, then build a system around it, on the software they already use. The system starts by only watching, earns more responsibility step by step, keeps proof of everything it does, and always stops for a person on the big decisions. - Lead with the work moving, not the technology. Our words are flow, stuck, and from start to finish. - Our idea sounds like a contradiction: software that runs on its own but starts with no freedom at all. Say it plainly and repeat it. - Talk about proof: tested on real cases, a record of every step, and a system the client owns. Never AI magic, never a confidence score. - It takes over the chasing, not the decisions. Say hours back and the work people were hired for, never productivity. ## The name. Falnor has two syllables. Fal is the motion: work moving forward. Nor is the north star: direction, and the point it is heading towards. Together they mean momentum with direction. Not just speed, but speed that is aimed. Work that flows because it knows where it is going. - Fal: Forward motion. Momentum and flow. Work moving from start to finish without anyone pushing it along. - Nor: North star. Direction and guidance: the rules that decide, the checks, and the person it stops for. Speed only helps when it is aimed. Momentum with direction. Speed that is aimed, not just speed. ## Colour. Deep Ocean carries authority. Navy and warm white do most of the work. Amber is kept for what matters most: the decision, the button, the task that is waiting. No purple “AI” colours, no gradients, no mascots. - Deep Ocean #20374D: Primary. Headings, dark bands, the frame - Electric Amber #C98917: Accent. CTAs, the waiting item, the accent word - Soft Amber #E7B65C: Amber on dark surfaces, hover - Pale Blue #C3D4E6: Secondary surfaces, chart fills - Steel Blue #7A8C9D: Supporting UI, citations, icons - Slate Navy #4F6578: Body copy on light surfaces - Warm White #F8F6F2: Page background, light bands - Graphite #1D232B: Dense UI text, mono values ## Typography. Clash Display for headings and the accent word, semibold only. Supreme for body text and on screen. A fixed-width font for references, money and times, so a number always looks like a number. - Clash Display: Headings and the accent word. Semibold. Tight tracking, balanced wrap - Supreme: Body copy, interface, navigation. Regular. 15 to 17px, 1.6 leading - Monospace: References, money, times. Berkeley Mono in the product. System font on the web ## Words. Use these words the same way every time. Consistency is what makes the story believable. Write in plain British English, in short sentences. We do not say seamless, unlock, powerful, magic, next-gen, copilot, assistant, blocked, confidence score or productivity. - Flow: The whole journey from start to finish. Not one step, not one connection between apps. - Stuck: What work does when it waits for a detail, a sign-off or a person. Never “fails”. - Chasing: What experts do by hand today. The thing we take away. Never blame “your team”. - Task: One piece of work: a claim, an invoice, a customer request. - Waiting: A step that needs a person. Never “blocked”. - Check: Where the system stops and asks a named person before anything that cannot be undone. - Watching: How every system starts: suggesting what it would do while a person still does the work. - Earned: The system is allowed to do more only once results show it is ready. - Record: What it looked at, what it changed and the rule that allowed it. Never a score. - Yours: The system, its rules, its record and its test cases belong to the client from day one. ## Next to another brand. - Logo clearance: Keep clear space around the lockup equal to the height of the mark. Do not crowd it with text, borders or competing marks. - Colour discipline: Deep Ocean for authority, Electric Amber sparingly for the one thing that matters. No off-palette gradients unless approved for a co-branded asset. - Voice: Plain British English, short sentences and everyday words. Every number with its source and date. No em dashes. Say what the system does, and let the reader picture their own work. ## Before you publish. - Use approved logo files only: Download the lockup and mark from this page. Do not recreate the wordmark, stretch assets or change logo colours. - Attribute Falnor correctly: First mention: Falnor. You may include the tagline on partner materials. Link to falnor.io when publishing digitally. - Ask before co-branding: For event logos, slide decks or joint announcements, get in touch through our contact page so we can check placement and contrast. ## For press and analysts. Three stories we can back with a named source. - Work does not fail. It stops.: The hidden cost of office work is the chasing: 58% of the working day goes on work about work (Asana, 2023). What changes when a system takes that off people’s desks. - Software that starts with no freedom: Why every Falnor system starts by only watching, earns more responsibility on results, and steps back automatically after a mistake. Trust built on a track record, not switched on. - The supplier you can walk away from: The system, its rules, its record and its test cases are handed over on day one. What changes when a client can replace the supplier and keep the system. # Contact (https://falnor.io/contact) ## Book a 30-minute discovery call Tell us about one part of your business where work gets stuck. You will get a straight answer on whether a system would help. If it would not, we will tell you on the call. - Show us where work gets stuck: Tell us where things wait, what you have already tried, and what good would look like for your team. - Choose a starting point: Mapping one process, building a first system, or fixing something that only half works today. We will share examples from similar businesses. - Walk us through your software: Which tools the work passes through, what needs a person, and what your managers or auditors will want to see. - Leave with a clear next step: An honest view on fit, and what mapping would cover. The scope and price come after the map. No obligation. ### A good fit looks like - A clear process with someone in your team who owns it - A wish to fix the whole job, not run another trial - Your team staying in control until the system has earned trust ### Before you book - What should I bring to the call? One process where work gets stuck, the software it passes through, and who owns it. A rough picture is enough. Mapping fills in the rest. - What if a system would not help us? We will tell you on the call. The aim is a straight answer on fit, not a sale. - Does the call commit us to anything? No. The call carries no obligation. If you go ahead, the scope, timeline and price are agreed in writing after the map and before anything is built. - Which countries and time zones do you work in? We work with businesses in the US, the UK and Europe. If none of the slots suit your time zone, send a message and we will find one. Prefer to send a message? Tell us about the process, the software you use and your timing. We reply within one working day. # AI automation agency (https://falnor.io/solutions/ai-automation-agency) ## An AI automation agency that builds the whole system, not a pile of automations An AI automation agency designs and builds the software systems that take repetitive operational work off a team. Falnor does this for B2B operations: we map how one part of your operation really runs, build a system around it on the software you already use, and keep a person in charge of every important decision. We work with businesses in the US, the UK and Europe. Not sure how to pick one? See [how to choose an AI automation agency](/guides/how-to-choose-an-ai-automation-agency). ## Why single automations stall A trigger that moves one field from one tool to another is easy to build and easy to outgrow. The work that actually eats your week is the exceptions, the unwritten rules and the sign-offs, and those are what a quick automation skips. See how that compares with Zapier, Make or n8n. - Exceptions land back on your team, so the chasing never really stops. - Nobody can say why the automation did what it did, so nobody trusts it. - It breaks quietly when a tool or an AI model underneath changes. ## How an engagement runs Every engagement follows the same four steps, whichever part of the business we start with. - Map the work: We write down how the work really runs: where each detail comes from, the rule at each step, who decides and what gets updated. You keep the map whether or not we build. - Build the system: We build one system around that map on the tools you already have, rather than asking you to move to a new platform. - Start in shadow: The system first only reads your software and shows what it would do. It earns the right to act on one task at a time, once it is right at least 9 times in 10 on your real cases. - Hand it over: You own the system, its rules and its full record from day one, and it keeps running without us. ## A person decides what matters Anything important, unusual or hard to undo waits for a named person to approve it. Every step is recorded: what the system looked at, which rule it followed and who approved it. This is the idea behind human-in-the-loop AI. Works with: Xero, Sage, Salesforce, HubSpot, SAP, NetSuite, Microsoft Dynamics, Zendesk. We work with businesses in the US, the UK and Europe. ## Questions people ask ### What does an AI automation agency do? It designs and builds the systems that take repetitive operational work off your team. At Falnor that means mapping the work first, then building one system around it on your own software, with people approving the important decisions. ### How is Falnor different from an agency that builds Zapier or Make flows? Those flows repeat one step you already understand. We build the whole system around the work: the unusual cases, the unwritten rules, the points where a person decides, and a record of everything. ### Which countries do you work in? We work with businesses in the US, the UK and Europe. The system runs on your own systems, so your data stays where it is. ### How much does it cost? There is no price list. We map your work first, and you get the scope, timeline and price in writing before anything is built. # Business process automation (https://falnor.io/solutions/business-process-automation) ## Business process automation built around how your work really runs Business process automation only helps when it matches the process you actually run, not the one in the slide deck. Falnor maps the real path first, then builds a system on the software you already use that moves each item from request to closed record. ## Where processes stop An invoice, a claim, a new customer. The real work takes minutes. Then it sits and waits for a missing detail, a sign-off, or someone to remember it. Your best people keep it moving by hand. - Work waits in inboxes and spreadsheets between tools. - The rules live in people’s heads, so every exception is a question. - Reporting is a manual job because nothing records why a step happened. ## What the system does A Falnor system owns the job from the request coming in to the record being closed. For finance work, see accounts payable automation. For incoming requests, see customer service automation. - Reads everything first: Every connected tool is read-only until the map of your process is trusted. Evidence is where each detail was read from, never a confidence score. - Follows clear rules: Rules you have approved decide, not guesswork. When no rule fits, the item goes to a person. - Closes the item: Not a draft or a summary: the record updated, the customer answered, the exception cleared and the next system told. - Keeps the record: Each decision is logged with what it looked at, the rule it followed and who approved it. ## Automation that earns trust The system starts by only reading. It takes over one task at a time, once it is right at least 9 times in 10 on your real cases, and pauses safely before anything that cannot be undone. Works with: Xero, Sage, Salesforce, HubSpot, SAP, NetSuite, Microsoft Dynamics, Zendesk. We work with businesses in the US, the UK and Europe. ## Questions people ask ### What is business process automation? It is using software to carry out the steps of a recurring business process, such as approving an invoice or routing a customer request, with less manual chasing. ### Is this the same as workflow automation? They overlap. Workflow automation often means a chain of triggers. We build the whole system around the process, including the exceptions, the checks and the record. ### Do we need to replace our current software? No. The system is built on the tools you already use, such as Xero, Salesforce, HubSpot or NetSuite. ### Our process is not written down. Can you still automate it? Yes. Mapping writes it down first. In most businesses it lives in people’s heads, and you keep the map either way. # Accounts payable and invoice automation (https://falnor.io/solutions/accounts-payable-automation) ## Accounts payable and invoice automation that stops at a person, not a queue Invoices wait for a signature while your team chases details. Falnor builds a system on your finance software that reads each invoice, checks it against your rules, and routes anything unusual to a named approver before money moves. ## Where invoices get stuck Most invoice work is simple. The delay comes from the invoices that are not: a missing purchase order, a changed bank detail, a total that does not match, an approver who is away. - Details are re-typed between email, the finance system and spreadsheets. - Approvals wait on one person remembering to look. - Month-end turns into a hunt for what is still open. ## What the system does with each invoice It follows the path your best AP person already follows, and writes down every step. It connects to Xero and NetSuite. - Reads the invoice: It pulls the supplier, amounts, dates and references, and checks them against what your finance system already holds. - Applies your rules: Matching, limits and coding follow rules you approved. If nothing matches, it stops and says why. - Asks the right person: Anything unusual, large or hard to undo waits for a named approver, with the evidence beside it. - Updates the record: Once approved, it updates your finance system and tells the next step, and logs who approved what. ## No payment without a person The system begins in shadow, reading and proposing only. It takes over routine invoices one type at a time once it is right at least 9 times in 10 on your own invoices, and everything else keeps going to a human. Works with: Xero, Sage, NetSuite, SAP, Microsoft Dynamics. We work with businesses in the US, the UK and Europe. ## Questions people ask ### What is accounts payable automation? It is using software to handle the repeatable parts of paying suppliers: capturing invoice data, matching it to orders, routing approvals and posting to the ledger. ### Will it replace my finance team? No. It takes over the chasing, not the decisions. Your team keeps the judgement calls. ### Does it work with Xero, Sage or NetSuite? We build on the finance software you already have. Xero, Sage, NetSuite, SAP and Microsoft Dynamics are common starting points. ### How do you stop a wrong payment? Anything unusual, large or hard to undo waits for a named approver. Every decision is recorded with the rule it followed. # Human in the loop AI (https://falnor.io/solutions/human-in-the-loop-ai) ## Human-in-the-loop AI agents that pause for a person before anything important AI agents are useful in operations only if you can trust what they do. A Falnor system looks before it acts, follows rules you approved, stops for a named person on anything important or hard to undo, and records every step. ## The trust problem with agents An agent that acts on its own is fast until it is wrong in a way nobody sees. In finance, claims and customer work, that is not a risk most teams can take. - No clear line between what the agent may do alone and what needs a person. - No record of why a decision was made. - Behaviour that changes silently when the AI model underneath changes. ## How the human stays in the loop These are the principles every Falnor system is built on, whichever AI sits underneath. - Look before acting: The system starts read-only and shows what it would do before it does anything. - Clear rules decide: Rules you approved decide the outcome. The AI helps read and prepare, it does not make up policy. - Pause before big steps: Important, unusual or irreversible steps wait for a named approver, with the evidence beside them. - Proof, not guesswork: Every decision records what was looked at, which rule applied and who approved it. ## Autonomy is earned, one task at a time A task moves from proposed to automatic only after the system is right at least 9 times in 10 on your real cases. If the AI underneath changes, we re-test on your cases before anything changes. Works with: Xero, Salesforce, HubSpot, NetSuite, Zendesk. We work with businesses in the US, the UK and Europe. ## Questions people ask ### What does human in the loop mean? It means a person reviews or approves the steps that matter. The system does the routine work and hands the important decisions to a named human. ### Does a person have to approve everything? No. Routine, well-understood items can run automatically once the system has proven itself on your cases. Anything unusual or hard to undo still waits for a person. ### What if the AI model changes? We re-test on your own cases before anything changes in production, so behaviour does not shift without you knowing. ### Where does our data go? The system runs on your own systems and reads from them, so your records stay where they are. # Customer service and request automation (https://falnor.io/solutions/customer-service-automation) ## Customer service automation that answers the routine and hands over the rest Customers wait for a reply while your team looks things up across three tools. Falnor builds a system on your helpdesk and CRM that reads each request, finds the facts, drafts or sends the routine answer, and passes anything unusual to the right person with the context attached. ## Where customer requests wait The answer is usually in your systems already. The delay is finding it, checking the policy, and getting the right person to sign off. It works with HubSpot and Salesforce. - Requests sit in a shared inbox until someone triages them. - Agents switch between helpdesk, CRM and finance tools to answer one question. - Customers chase for updates because nobody can see where a request has got to. ## What the system does with each request It works the request the way your best agent would, and shows its working. - Triage: It reads the request and sorts it by what the customer needs and how urgent it is. - Gather the facts: It pulls the account, order and history from your own systems, read-only at first. - Answer or escalate: Routine requests get an answer under your rules. Anything unusual goes to a named person with the evidence beside it. - Close the loop: It updates the record, tells the customer and logs what happened. ## Your team keeps the judgement calls Refunds, exceptions and anything sensitive wait for a person. The system takes over the chasing, not the decisions, and starts in shadow before it sends anything. Works with: Zendesk, HubSpot, Salesforce, Microsoft Dynamics. We work with businesses in the US, the UK and Europe. ## Questions people ask ### Is this a chatbot? No. It is a system that works inside your helpdesk and CRM to triage, gather facts and prepare or send routine answers, with a person approving the rest. ### Will customers know they are talking to automation? That is your call. We build to your policy and tone, and nothing important is sent without the approval rules you set. ### Which tools does it work with? We build on the tools you already use, such as Zendesk, HubSpot and Salesforce. ### How long does it take to start? It starts with a 30-minute discovery call, then mapping. Timeline and price are agreed in writing after mapping. # Claims processing automation (https://falnor.io/solutions/claims-processing-automation) ## Claims processing automation with a person on every decision that matters A claim takes minutes to assess and days to move. Falnor builds a system that gathers the documents, checks the claim against your own rules, and routes anything unusual to a named handler, keeping a full record of why each decision was made. ## Why claims wait Most delay is administrative: missing documents, details spread across systems, and handoffs between people who each hold part of the picture. - Handlers re-key the same facts into several systems. - Missing information is found late, then chased by email. - Audit trails are rebuilt by hand when someone asks why. ## What the system does with each claim It prepares the claim so the handler decides faster, and records everything. - Collect and check: It reads the claim and documents, spots what is missing and asks for it early. - Check against your rules: Coverage, limits and thresholds follow rules you approved, with the source of each fact shown. - Hand over the unusual: Large, unusual or disputed claims go to a named handler with the evidence attached. - Update and record: Once decided, it updates the claims system and logs the rule followed and who approved it. ## Decisions stay with your handlers The system starts in shadow, only reading and proposing. It takes routine steps over one at a time once it is right at least 9 times in 10 on your real claims. Anything hard to undo waits for a person. Works with: Salesforce, Microsoft Dynamics, SAP, Zendesk. We work with businesses in the US, the UK and Europe. ## Questions people ask ### What is claims processing automation? It is using software to handle the repeatable parts of a claim: collecting documents, checking details, applying rules and routing for a decision. ### Does it approve or deny claims on its own? Not by default. Decisions that matter wait for a named person, and any automatic step is limited to cases the system has proven right on your data. ### Is it suitable for regulated work? The system records what it looked at, which rule it followed and who approved it, and runs on your own systems so data stays where it is. We build with GDPR in mind. ### Which claims systems does it work with? We build on the tools you already use. Tell us what you run on the discovery call. # Business automation consulting (https://falnor.io/solutions/business-automation-consulting) ## Business automation consulting that ends with a working system, not a slide deck Falnor maps how one part of your business really runs, tells you plainly whether automation would help, and then builds the system on the software you already use. We work with businesses in the US, the UK and Europe. ## What automation consulting often leaves behind A common result of consulting is a report: a ranked list of processes that could be automated. Someone else then has to build it, and the details that matter, the exceptions and the approvals, get lost on the way. If you are weighing an agency against building the skills yourself, see AI automation agency vs in-house. - A gap between the recommendation and the thing that actually runs. - Rules that live in people’s heads and never make it onto paper. - No record of why a step happened, so nobody trusts the result. ## How an engagement works It starts with a 30-minute discovery call and an honest answer on fit. - Discovery call: You show us where work gets stuck and which tools it passes through. If a system would not help, we tell you on the call. - Map the work: We write down how the work really runs: where each detail comes from, the rule at each step, who decides and what gets updated. You keep the map whether or not we build. - Agree scope in writing: The scope, timeline and price are agreed in writing after the map and before anything is built. There is no price list. - Build and prove it: The system starts in shadow, only reading your software. It takes over one task at a time, once it is right at least 9 times in 10 on your real cases. - Hand it over: You own the system, its rules and its full record from day one, and it keeps running without us. ## A straight answer on fit Not every process should be automated. Some are too rare or too risky. Anything important, unusual or hard to undo waits for a named person, and every step is recorded. Works with: Xero, Sage, Salesforce, HubSpot, SAP, NetSuite, Microsoft Dynamics, Zendesk. We work with businesses in the US, the UK and Europe. ## Questions people ask ### What does a business automation consultant do? A consultant works out which parts of your operation could run with less manual effort and how. At Falnor that includes building the system, so the recommendation does not stop at a report. ### How is this different from a large consultancy? We map one part of your business and then build the system around it, rather than producing a report for someone else to build from. ### How much does it cost? There is no price list. We map your work first, and you get the scope, timeline and price in writing before anything is built. ### Do we need to replace our software? No. The system is built on the tools you already use, such as Xero, Salesforce, HubSpot or NetSuite. ### Which countries do you work in? We work with businesses in the US, the UK and Europe. The system runs on your own systems, so your data stays where it is. # Salesforce automation (https://falnor.io/integrations/salesforce) ## Salesforce automation built around your process, not just your flows Salesforce can automate a lot, until the work crosses into your finance system, your inbox or a spreadsheet. Falnor builds a system on top of Salesforce and the tools around it that owns the job end to end. ## Where Salesforce work stops Flows handle steps inside Salesforce well. The gaps appear at the edges: data that has to match another system, approvals outside the CRM, and exceptions nobody wrote a rule for. - Records are updated by hand after something happens in another tool. - Approvals live in email, not in the record. - Nobody can see why a field changed. ## What we build around Salesforce The system connects to Salesforce through its own permissions, read-only first. - Read first: It reads the records it needs and shows what it would change before it changes anything. - Follow your rules: Updates, routing and checks follow rules you approved, including the ones that cross into other systems. - Ask for approval: Important or unusual changes wait for a named person, with the evidence in view. - Keep the record: Each change is logged with the rule followed and who approved it. ## Your Salesforce, your rules We do not move you to a new platform. The system runs alongside Salesforce, you own its rules and its record, and it keeps running without us. Works with: Salesforce, Xero, NetSuite, HubSpot, Zendesk. We work with businesses in the US, the UK and Europe. ## Questions people ask ### Is this a Salesforce consultancy? No. We build the system around your process that uses Salesforce as one of its tools, rather than configuring Salesforce itself. ### Do you replace Salesforce Flow? No. Flows are good inside Salesforce. We cover the work that crosses systems and needs a person to approve. ### Does it write to Salesforce straight away? No. It begins read-only and earns the right to write, one task at a time. # HubSpot automation (https://falnor.io/integrations/hubspot) ## HubSpot automation that carries the work past the workflow HubSpot workflows handle the steps inside HubSpot. Falnor builds the system around them: reading what happens in your finance, support and delivery tools, updating HubSpot, and asking a person when a decision is needed. ## Where HubSpot work stops A deal closes, a customer replies, a contract is signed, and then someone has to carry that into other tools by hand. - Deal and customer data is re-entered in finance and support tools. - Handoffs between sales, finance and delivery depend on a message. - Exceptions to the standard path have no home. ## What we build around HubSpot The system connects to HubSpot through its own permissions, read-only first. - Read first: It reads the deals, contacts and tickets it needs and shows what it would do. - Follow your rules: Updates, routing and checks follow rules you approved. - Pause for a person: Anything unusual, large or hard to undo waits for a named approver. - Keep the record: Each change is logged with the rule followed and who approved it. ## You keep HubSpot as it is We build on the HubSpot you already run. You own the system, its rules and its full record from day one. Works with: HubSpot, Xero, Salesforce, Zendesk. We work with businesses in the US, the UK and Europe. ## Questions people ask ### Do you replace HubSpot workflows? No. Workflows are good for steps inside HubSpot. We handle the work that spans tools and needs a person to decide. ### Can it update our finance system from HubSpot deals? That is a common use. It reads the closed deal, checks it against your rules and updates the finance system, with approval where you want it. ### Does it work with HubSpot Service Hub? We build on whatever HubSpot setup you run. Tell us what you use on the discovery call. # NetSuite automation (https://falnor.io/integrations/netsuite) ## NetSuite automation for finance and operations teams NetSuite holds the numbers, but the work around it still moves by email and spreadsheet. Falnor builds a system on NetSuite and the tools around it that handles the routine and asks a person about the rest. ## Where NetSuite work stops Records get into NetSuite late, wrong or by hand, and the people who could fix them are busy chasing the next item. - Invoices, orders and journal details are keyed in from other systems. - Approvals sit in email threads. - Reconciling what happened and why takes a day each month. ## What we build around NetSuite The system connects to NetSuite through its own roles and permissions, read-only first. - Read first: It reads the records it needs and shows what it would post before posting anything. - Follow your rules: Matching, limits and coding follow rules you approved. - Pause for a person: Anything unusual, large or hard to undo waits for a named approver. - Keep the record: Every posting is logged with the rule followed and who approved it. ## Your NetSuite stays the system of record We do not move your data. The system runs on your own systems and reads from them, and you own its rules and record. Works with: NetSuite, Salesforce, Xero, SAP. We work with businesses in the US, the UK and Europe. ## Questions people ask ### Is this a NetSuite implementation partner? No. We build the system around your process that uses NetSuite as one of its tools. We do not configure or resell NetSuite. ### Can it help with accounts payable? Yes. Invoice capture, matching and approval routing are a common starting point. See our accounts payable automation page. ### Does it post to NetSuite on its own? Not at first. It begins read-only and earns the right to post, one task at a time. # Xero automation (https://falnor.io/integrations/xero) ## Xero automation for invoices, approvals and the work around them Xero keeps the books. The chasing happens around it: invoices waiting for a signature, details to match, approvals by message. Falnor builds a system on Xero that carries that work and asks a person only when a decision is needed. ## Where Xero work stops The bookkeeping is tidy once it is in Xero. Getting it there is the slow part. The same approach is described for accounts payable automation and business process automation. - Bills arrive by email and are keyed in by hand. - Approvals wait on one person. - Month-end is a hunt for what is still open. ## What we build around Xero The system connects to Xero through its own permissions, read-only first. - Read first: It reads the bills and contacts it needs and shows what it would do. - Follow your rules: Matching, limits and coding follow rules you approved. - Pause for a person: Unusual or large items wait for a named approver. - Keep the record: Each posting is logged with the rule followed and who approved it. ## Your books, your approvals You own the system, its rules and its full record from day one, and it keeps running without us. Works with: Xero, HubSpot, Salesforce. We work with businesses in the US, the UK and Europe. ## Questions people ask ### Is Falnor a Xero partner? We build on top of Xero as one of the tools you already use. We do not claim any partnership or certification. ### Can it handle supplier invoices? Yes. Reading, matching and routing for approval is a common starting point. See our accounts payable automation page. ### Will it post to Xero automatically? Not at first. It starts read-only and earns the right to act, one task at a time. # Zendesk automation (https://falnor.io/integrations/zendesk) ## Zendesk automation for customer requests that need facts from other tools Zendesk holds the ticket. The answer often lives somewhere else: the CRM, the finance system, an order record. Falnor builds a system around Zendesk that reads the request, fetches those facts, prepares the reply and asks a person before anything sensitive goes out. ## Where Zendesk work stops Triage rules and macros handle the routine. The slow part is the agent opening three other tools to find out what is true. The same approach is described for customer service automation and, for the CRM side, HubSpot automation. - Agents look up account, invoice or order details in other tools before they can reply. - Refunds, credits and exceptions wait on a manager with no evidence in front of them. - Ticket history and customer records in the CRM drift apart. ## What we build around Zendesk The system connects to Zendesk through its own permissions, read-only first. - Read first: It reads incoming tickets and the records it needs from your other tools, and shows what it would send before it sends anything. - Follow your rules: Routing, answers from your approved sources and limits on refunds or credits follow rules you wrote. - Pause for a person: Sensitive replies, above-limit refunds and unclear cases wait for a named person, with the ticket history and evidence beside the decision. - Keep the record: Each action is logged with what it looked at, the rule it followed and who approved it. ## Your Zendesk, your rules We do not replace your helpdesk or your agents. The system runs alongside Zendesk, you own its rules and its full record from day one, and it keeps running without us. Works with: Zendesk, HubSpot, Salesforce, Xero. We work with businesses in the US, the UK and Europe. ## Questions people ask ### Is Falnor a Zendesk implementation partner? No. We build on top of Zendesk as one of the tools you already use. We do not claim any partnership or certification. ### Can it connect Zendesk tickets to HubSpot records? Yes. It reads the ticket and the matching HubSpot contact, company or deal, prepares the reply or update, and asks a person where you want approval. ### Will it reply to customers on its own? Not at first. It starts read-only, then drafts, and earns the right to send routine replies one task at a time. Anything sensitive keeps waiting for a person. ### How is this different from Zendesk macros and triggers? Macros and triggers act on fields inside Zendesk. We cover the work that needs facts from other systems and a person to approve the important steps. # Sage automation (https://falnor.io/integrations/sage) ## Sage automation for invoices where every approval has a name Sage keeps the ledger. The slow part is getting an invoice approved by the right person and posted correctly. Falnor builds a system around Sage that reads each invoice, checks it against your rules and puts a named approver on anything that needs one. ## Where Sage work stops Once a bill is coded and posted, Sage does its job. Getting it there is where the time goes. See accounts payable automation for the wider approach and NetSuite automation if you run both. - Invoices arrive by email or PDF and are keyed in by hand. - Approvals depend on who is in the office and who remembers. - Nobody can show later who approved an invoice and on what basis. ## What we build around Sage The system connects to Sage through its own permissions, read-only first. - Read first: It reads the invoices, suppliers and purchase orders it needs and shows what it would do. - Follow your rules: Matching, coding and approval limits follow rules you approved, including who approves what. - Pause for a person: Unusual, large or unmatched invoices wait for a named approver, and the system chases them if they sit. - Keep the record: Each posting is logged with the rule followed and who approved it, so the answer to "who signed this off" is one click away. ## Your ledger, your approvers We do not move you off Sage. You own the system, its rules and its full record from day one, and it keeps running without us. Works with: Sage, Xero, NetSuite, HubSpot. We work with businesses in the US, the UK and Europe. ## Questions people ask ### Which Sage products do you work with? We build on the Sage product you already run. Tell us which one on the discovery call and we will say plainly whether it is a good fit. ### Is Falnor a Sage partner? We build on top of Sage as one of the tools you already use. We do not claim any partnership or certification. ### Can every invoice still need a named approver? Yes. That is the design. The system routes each invoice to a named person by your rules, chases it, and records who approved it. ### Will it post to Sage automatically? Not at first. It starts read-only and earns the right to post, one task at a time, once it is right at least 9 times in 10 on your real invoices. # AI agents for business (https://falnor.io/guides/ai-agents-for-business) ## AI agents for business: what they do, where they go wrong, and who should approve An AI agent is software that reads information, picks a next step and acts on it, often across several tools. That makes it useful for everyday operations work, and risky if nobody is watching. This guide covers the uses, the failure modes and how to set one up so a person stays in charge. ## What an AI agent is An agent takes in information such as an email, an invoice or a support ticket. It decides what should happen next, using rules and a language model, and then does it in the software your team already uses: updating a record, drafting a reply, creating a task or asking someone to approve. It helps to separate three things that often get mixed up. - A chatbot answers a question. It does not usually change anything in your systems. - A fixed automation follows one path you set up in advance, such as "when a form is submitted, add a row". It cannot cope with a case you did not plan for. - An agent handles cases that vary. It reads what arrived, works out which path applies and takes the steps, which is why it needs limits. For a side-by-side with older tools, see RPA vs AI agents. ## Where businesses use AI agents The best fits are jobs where the real work takes minutes and then the item waits for a detail, a check or a sign-off. - Finance: reading supplier invoices, matching them to orders and routing approvals. See accounts payable automation. - Claims and cases: collecting documents, spotting what is missing and preparing the file for a handler. See claims processing automation. - Customer requests: sorting incoming requests, pulling the account history and drafting the routine answer. See customer service automation. - Operations between tools: keeping the CRM, the finance system and the helpdesk consistent without re-typing. - Onboarding: gathering documents, checking them and telling the next team when a customer is ready. ## Where AI agents go wrong Most problems come from an agent that is given too much room, not from the technology being useless. - It acts confidently on bad or incomplete data. - It fills a gap with a plausible guess when it should have stopped and asked. - It takes a step that cannot be undone, such as sending a payment or emailing a customer. - Its behaviour shifts quietly when the AI model underneath is updated. - Nobody can say afterwards why it did what it did. ## How to keep a person in charge These are the steps we use when we build a system, whichever AI sits underneath. The approval step has its own guide: human-in-the-loop automation. - Start read-only: Let the agent read your software and show what it would do before it is allowed to change anything. - Write the rules down: Rules you have approved should decide the outcome. The AI helps read and prepare the item. It should not invent policy. - Decide what needs a person: Anything important, unusual or hard to undo waits for a named approver, with the evidence shown next to the decision. - Keep a record: Log what it looked at, which rule it followed and who approved it, so any decision can be explained later. - Earn autonomy one task at a time: Let a task run on its own only after it has been right at least 9 times in 10 on your real cases. Re-test on those cases whenever the model underneath changes. ## Questions to ask before you buy or build - What can it read, and what can it change? - What does it do when it is unsure? - Who approves, and how quickly can they respond? - Can I see why it made a decision? - Who owns the rules and the record if we stop working with you? - Does it run on our own systems, so our data stays where it is? ## Questions people ask ### What is the difference between an AI agent and a chatbot? A chatbot answers questions. An agent also takes actions in your software, such as updating a record or routing an approval, which is why it needs rules and a person approving the important steps. ### Are AI agents safe to use for finance work? They can be, with the right limits. Start read-only, write the rules down, and make anything unusual or hard to undo wait for a named person. Every step should be recorded. ### Will AI agents replace my team? A well-built system takes over the chasing, not the decisions. Your people keep the judgement calls, and the system learns from them. ### How much does an AI agent system cost? There is no price list. Every business works differently, so we map the work first and agree scope, timeline and price in writing before anything is built. ### How do I start? Pick one part of the business where work gets stuck and book a 30-minute discovery call. You get a straight answer on whether a system would help. # Human-in-the-loop automation (https://falnor.io/guides/human-in-the-loop-automation) ## What is human-in-the-loop automation, and where to put the person Human-in-the-loop automation means software does the routine work while a person reviews or approves the steps that matter. It gives you the speed of automation without handing over decisions you cannot take back. ## What human in the loop means There are three common setups, and the names are easy to confuse. - Human in the loop: the system prepares the action and waits for a person to approve it before anything happens. - Human on the loop: the system acts on its own while a person monitors it and can step in or reverse a decision. - Human out of the loop: the system acts with no one watching. In business operations this is rare, and sensible only for low-stakes steps that are easy to undo. ## When to use each one Many systems use all three, depending on the step. The skill is deciding which step gets which. - Use in the loop for payments, changes to bank details, claim decisions, customer refunds and anything hard to undo. - Use on the loop for routine items the system has already proven it handles correctly, such as matching an invoice that agrees exactly with its purchase order. - Treat out of the loop as the exception, and only for steps where a mistake is cheap and reversible. ## Examples from finance, claims and customer work - Supplier invoices: An invoice that matches its order within your limits can move on automatically. One with an unusual amount or a changed bank detail goes to a named approver with the evidence beside it. See accounts payable automation. - Claims: Checking that the documents are complete can run on its own. A large, unusual or disputed claim goes to a handler to decide. See claims processing automation. - Customer requests: A routine answer can be drafted from the account history. A refund or an exception waits for a person. See customer service automation. ## How to design the approval step - Decide what triggers a stop: Use clear rules, such as an amount over a limit or a field that changed, rather than a feeling that something looks odd. - Show the evidence next to the decision: An approver who has to hunt for context will either slow down or approve without checking. - Name the approver: Send it to a person, not a shared inbox that nobody owns. Say who covers when they are away. - Record every approval: Keep what the system looked at, which rule applied and who said yes, so you can explain it later to a manager or an auditor. - Review and loosen over time: Look at what gets approved. If a type of item is always approved, it may be ready to run on its own. ## Common mistakes - Asking for approval on everything, so approvers click through without reading. - Showing the decision without the evidence behind it. - Routing approvals to a team inbox with no owner. - Never revisiting the rules, so the people stay in the loop long after the system has earned trust. ## Questions people ask ### What is the difference between human in the loop and human on the loop? With a person in the loop, the system waits for approval before it acts. With a person on the loop, it acts first and a person monitors and can step in. ### Does human approval slow everything down? Only the items that need it. Routine, well-understood items run on their own, and the unusual ones wait for a person with the evidence in front of them. ### How do you decide what needs approval? We map how the work really runs, then set rules for what counts as important, unusual or hard to undo. A system starts read-only and earns the right to act on a task once it is right at least 9 times in 10 on your real cases. ### Is human oversight required by regulation? Some rules call for it. The EU AI Act, for example, expects human oversight of high-risk AI systems. We are not a law firm, so check what applies to your business with your own adviser. # How to choose an AI automation agency (https://falnor.io/guides/how-to-choose-an-ai-automation-agency) ## How to choose an AI automation agency: what they do and what to ask An AI automation agency designs and builds the software systems that take repetitive operational work, such as invoices, claims and customer requests, off a team. Choosing one comes down to whether they understand how your work really runs, who stays in charge of the decisions, and who owns the result. This guide covers what they do, when to hire one and the questions to ask. ## What an AI automation agency does An agency takes a part of your operation that is slow or error-prone, works out how it runs today, and builds software around it. The software reads what comes in, applies your rules, updates the tools you already use, and asks a person when a case needs a decision. Agencies vary a lot in what they deliver. Three kinds are common. - Workflow builders connect tools with triggers and prompts, usually in a no-code platform. They are quick to start and fine for simple, stable steps. See Zapier, Make or n8n versus custom AI workflows. - Consultancies advise on strategy and tools and often leave the building to you. See business automation consulting. - System builders map one part of the operation, build a whole system around it, and hand it over with its rules and record. Falnor works this way. ## When to hire an agency An agency makes sense when the work is real but you cannot spare the people or the time to build it properly. - The work has exceptions and unwritten rules, so a single trigger would break on the first unusual case. - Your team spends hours a week chasing, re-typing or checking the same things. - You want an outside team to map the work, because the people who do it every day have stopped seeing the steps. - You do not have a developer who can own an AI system after it is built. ## When not to hire one - The step is simple and stable, such as copying a form into a spreadsheet. A self-serve tool does that well and costs little. - You already have an engineering team with time to build and maintain it. See AI automation agency versus in-house. - You cannot yet describe the process. Start by writing down how it runs now. A good agency will do that with you, but you should know it is the first job. ## Questions to ask before you sign - How do you learn how our work really runs?: Look for a mapping step before any building: where each detail comes from, the rule at each step, who decides and what gets updated. Ask whether you keep that map if the project stops. - Who approves the important decisions?: Anything unusual or hard to undo, such as a payment, a changed bank detail or a refund, should wait for a named person with the evidence in front of them. See human-in-the-loop automation. - What does it do when it is unsure?: It should stop and ask. A system that fills the gap with a confident guess is the main way these projects go wrong. - How does it earn the right to act on its own?: Ask whether it starts read-only and shows what it would do first. Autonomy should be granted one task at a time, after it has been right on your real cases. - Can I see why it made a decision?: You want a record of what it looked at, which rule applied and who approved, so any decision can be explained later to a manager or an auditor. - Who owns the system, the rules and the record?: You should, from day one. Ask what happens if you stop working with them, and whether it keeps running without them. - Does it run on our own software and data?: Building on the tools you already use avoids a platform migration and keeps your data where it is. - How is the price set?: Be wary of a quote before anyone has looked at the work. Scope, timeline and price should be agreed in writing after the work has been mapped. ## Red flags - They propose building a bespoke AI system before they have asked how the work runs. - The demo shows the happy path and nothing about exceptions. - Nobody can explain what the system does when it is unsure. - You would not own the rules, the record or the code. - Every answer is a platform you would have to move to. - There is no mention of a person approving anything. ## How Falnor works Falnor is an AI automation agency for B2B operations teams in the US, the UK and Europe. An engagement follows four steps: map the work, build one system on your own software, run it in shadow so it only shows what it would do, then hand it over. Details are on the AI automation agency page. If you want a straight answer on whether a system would help, you can book a 30-minute discovery call. ## Questions people ask ### What does an AI automation agency do? It designs and builds software that takes repetitive operational work off your team, such as reading invoices, preparing claims or sorting customer requests. A good one maps how the work runs first, builds around your existing tools, and keeps a person approving the important decisions. ### How do I choose an AI automation agency? Ask how they learn how your work really runs, who approves important decisions, what the system does when it is unsure, who owns the rules and the record, and how the price is set. Be cautious of anyone who quotes before they have looked at the work. ### When should I hire an agency instead of using Zapier or Make? Use a self-serve tool for simple, stable steps. Hire an agency when the work has exceptions, unwritten rules and sign-offs that a single trigger would miss, and you want the whole system built and handed over. ### How much does an AI automation agency cost? Prices vary by agency and by scope. Falnor has no price list: we map your work first, and you get the scope, timeline and price in writing before anything is built. ### What should I prepare before talking to an agency? Pick one part of the business where work gets stuck, and note who does it today, which tools it touches and where it goes wrong. A rough description is enough to start. # AI automation agency vs in-house (https://falnor.io/compare/ai-automation-agency-vs-in-house) ## AI automation agency or in-house: which fits your business? Both can work. The difference is who maps the process, who builds and maintains it, and what happens when the person who built it leaves. This page puts the two side by side and says when each one is the better choice. ## The real question is who holds the knowledge Automating a process well takes three things: understanding how the work really runs, building something that connects to your software, and keeping it right as things change. An agency brings all three on day one. An in-house hire brings the deepest knowledge of your business over time, but only as much of the other two as they have learned. ## When an agency is the better choice - You have one or two processes to fix, not a standing programme of work. - Nobody on your team has the time or the experience to build and test automation. - You want a person to approve the important steps and a record of every decision, and need them designed in from the start. - You would rather see a working first system before committing to more. ## When hiring in-house is the better choice - Automation is central to your product or operations and there is enough work for a full-time person. - The processes involve information you cannot share outside the company. - You already have engineers who want to own the work. ## A middle route Many businesses start with an agency to build and prove the first system, then bring ownership in-house once there is enough to maintain. If you go that way, ask at the start for the rules written down, the record of past decisions and a handover plan. That keeps the work yours either way. ## How Falnor works We map the work with your team first, then build the first system on the tools you already use. A named person approves anything important, every step is recorded, and we agree scope, timeline and price in writing before anything is built. There is no price list because every business runs differently. Read more about our AI automation agency and business automation consulting services, or how we keep a person in the loop. ## Questions people ask ### Is an AI automation agency cheaper than hiring someone? It depends on how much work there is. A single system is usually less than a full-time hire. A standing programme of work may favour a hire. We agree the price in writing after mapping the work, so you can compare real numbers. ### Who owns the automation if I stop working with an agency? It should be you. Ask for the written rules, the record of decisions and access to the systems to be yours from the start, and for a handover plan in the agreement. ### Can I start with an agency and move in-house later? Yes, and many businesses do. Plan for it at the start by asking for clear documentation and a handover, so your own team can take the system on. ### How do I start with Falnor? Pick one part of the business where work gets stuck and book a 30-minute discovery call. You get a straight answer on whether a system would help. # Zapier, Make or n8n vs custom AI workflows (https://falnor.io/compare/zapier-make-n8n-vs-custom-ai-workflows) ## Zapier, Make or n8n versus a custom AI workflow Self-serve tools are a good answer for many jobs, and often the right place to start. A custom AI workflow earns its keep when the work varies, needs judgement, or needs a person to approve the important steps. Here is how to tell which one you need. ## Start with the simple tool if it fits If a task is "when a form comes in, add a row and notify someone", a self-serve tool is fast and inexpensive to try. Many businesses get real value from them, and there is no reason to build anything custom for a job like that. ## Signs you have outgrown fixed flows - You keep adding exceptions and the flow is hard to follow. - Items arrive in different shapes, such as invoices or emails that are not laid out the same way twice. - A mistake would cost money, upset a customer or break a rule. - Someone has to approve before the next step, and you need to show who and why. - Only one person understands how the flows work. ## What a custom AI workflow adds - It reads what arrived: An AI model reads the email, document or request and works out what it is. - Rules you approved decide: Your written rules choose the outcome. The AI prepares the item and does not make up policy. - A person approves the important steps: Anything unusual, large or hard to undo waits for a named approver, with the evidence shown next to the decision. See human-in-the-loop automation. - Everything is recorded: You can see what it looked at, which rule it followed and who signed off. ## They can work together A custom workflow does not have to replace the tools you already use. Simple hand-offs can stay where they are while the judgement-heavy part is built properly. This is what we do under business process automation. If your tool is an older screen-recording bot, see RPA vs AI agents. We start by mapping the work and are happy to say when a simple tool is all you need. ## Questions people ask ### Is Zapier good enough for my business? For simple, repeatable hand-offs between apps, often yes. For work that varies, needs judgement or needs a person to approve important steps, a custom workflow is usually a better fit. ### Can n8n or Make use AI? They can call AI models as one step in a flow. What a flow does not give you by itself is the design around the AI: rules, approvals, limits and a record. That part is what we build. ### Do I have to replace my existing automations? No. We build on the tools you already use and leave simple hand-offs alone. We focus on the part where work gets stuck. ### How do I find out which I need? Book a 30-minute discovery call and describe the process. You get a straight answer, including when a simple tool is enough. # RPA vs AI agents (https://falnor.io/compare/rpa-vs-ai-agents) ## RPA or AI agents: what is the difference and which do you need? Robotic process automation (RPA) repeats steps someone recorded. An AI agent reads each case and decides what to do next. They suit different jobs, and in either case the important decisions should still have a person behind them. ## When RPA is enough If the work is the same every time, such as copying a fixed set of fields between two systems, recorded steps are a sound and predictable answer. Add monitoring so that a failure is noticed quickly. ## When an AI agent fits better - The information arrives in different forms, such as supplier invoices in many layouts. - The next step depends on what the item says, not only where it came from. - People currently spend time reading, checking and chasing before a simple decision. ## Why both need a person behind them Neither technology should make decisions that are hard to undo on its own. With RPA the risk is a quiet failure. With an agent it is a confident wrong answer. In both cases, set rules you have approved, make important steps wait for a named person, and keep a record so you can explain what happened. Our guide to human-in-the-loop automation covers where to put the approval, and human-in-the-loop AI shows how we build it. ## Combining them Some systems use an agent to read and prepare each item and a fixed integration to carry out the final step. We start by mapping the work and choose the simplest tool that does the job safely. If you are comparing against simple app-to-app tools, see Zapier, Make or n8n vs a custom AI workflow. ## Questions people ask ### Is an AI agent a replacement for RPA? Not always. RPA is still a sound choice for work that is identical every time. Agents are a better fit where cases vary and someone has to read and decide. ### Are AI agents less reliable than RPA? They fail differently. RPA tends to stop when something changes. An agent can act on a wrong reading, which is why it needs rules, approvals and a record. ### Can I keep my RPA and add an agent? Yes. An agent can read and prepare an item and hand it to the existing step. We map the work first and keep what already works. ### How do I start? Pick one part of the business where work gets stuck and book a 30-minute discovery call. You get a straight answer on whether a system would help. # Human in the loop vs human on the loop (https://falnor.io/compare/human-in-the-loop-vs-human-on-the-loop) ## Human in the loop or human on the loop: which one belongs where? In the loop, a person approves the action before it happens. On the loop, the system acts and a person watches, ready to step in. Most business operations need both, chosen step by step by how much a mistake would cost and how easily it can be undone. ## The short answer Human in the loop puts the person before the action. Human on the loop puts the person above it. The first is a gate. The second is a watch. A system can use a gate for one step and a watch for the next. ## How to choose for each step - Ask what a mistake would cost. If the answer is money, a customer or a rule, put a person in the loop. - Ask whether the action can be undone. If it cannot, put a person in the loop. - Ask whether the evidence is clear and the rule is written down. If it is, and the step is cheap to reverse, a person on the loop is usually enough. - Start with a person in the loop everywhere. Move a step to on the loop only after the system has been right on your real cases for long enough that you trust it. ## Examples from finance, claims and customer work - Accounts payable: Matching an invoice to its purchase order within limits can run with a person on the loop. Approving an unusual invoice, a new supplier or a change of bank details stays in the loop. - Claims: Reading documents and checking them against the policy can run on the loop. Deciding, paying or declining a claim stays in the loop with a named person. - Customer requests: Sorting requests and answering from approved sources can run on the loop. Refunds above a limit, sensitive complaints and account changes stay in the loop. ## Where approvals fail A person in the loop only helps if they can disagree. That needs the evidence beside the decision, the authority to say no, and a queue that is staffed. If nobody is available, an important action should wait rather than go ahead on its own. ## How Falnor builds it Every Falnor system starts with a person in the loop and read-only access. It earns the right to act on its own, one task at a time, once it is right at least 9 times in 10 on your real cases. Anything important, unusual or hard to undo keeps waiting for a named person, and every step is recorded: what it looked at, which rule it followed and who approved. Read more in our guide to human-in-the-loop automation, our page on human-in-the-loop AI agents and our approach. ## Questions people ask ### What is the difference between human in the loop and human on the loop? Human in the loop means a person approves the action before it happens. Human on the loop means the system acts and a person monitors it and can step in. One is a gate, the other is a watch. ### Which is safer for business operations? Human in the loop is safer for payments, refunds, claim decisions and anything hard to undo, because nothing important happens without approval. Human on the loop suits routine, reversible work with clear limits. ### Can one system use both? Yes, and most should. A system can run routine steps on the loop and stop for a named person at the steps that matter. ### Should we start with a person in the loop or on the loop? Start in the loop. Move a step to on the loop only after the system has proved itself on your real cases and you can pause and undo it. ### How do I start with Falnor? Pick one part of the business where work gets stuck and book a 30-minute discovery call. You get a straight answer on whether a system would help. # Privacy policy (https://falnor.io/privacy) Updated September 2026. falnor.io is operated by Falnor Technologies LLC. We design and build systems that run inside the software our clients already use. This page explains what personal data this website collects, why, who processes it, and the rights you have over it, including under the UK and EU GDPR. ## What this policy covers This policy covers the marketing site at falnor.io, the contact form, and the discovery-call booking embedded on the contact page. Data processed inside a client engagement (workflow records, the links to the client’s software, and the record of actions) stays in the client’s own systems and is governed by the statement of work and security review for that engagement, not by this page. ## Data we collect - Contact form: your name, work email, company, and anything you write in the message field. We also record the time the form was opened and submitted, and a hashed form of your IP address, to screen out automated spam. - Discovery-call booking: the name, email, timezone and any notes you give Cal.com when you pick a time. Cal.com also sends us booking, reschedule and cancellation events so we can confirm by email. - Server logs: our hosting provider keeps standard request logs (IP address, browser type, pages requested) for security and error diagnosis. - Product analytics: PostHog records pages viewed, referral source, and interaction signals such as clicks, tied to a pseudonymous visitor ID, with session replays that mask all form inputs. Google Analytics records aggregated traffic and page-level usage. We use both only to understand how the marketing site is used, and neither runs on the admin area. - We do not run advertising or cross-site tracking scripts on this site, and we never collect payment card details here. ## Why we use it - To reply to your enquiry and schedule or confirm a discovery call. - To send the booking confirmation, reschedule and cancellation emails for a call you booked. - To keep a record of who we have spoken to, for reasonable business record-keeping. - To detect and block spam and abuse of the contact form. Our lawful bases are your consent (when you submit the form or book a call) and our legitimate interest in responding to enquiries and protecting the site. ## Who processes it We do not sell or rent personal data. It is shared only with the processors below, and with legal advisers or authorities where the law requires it. - Convex: stores contact form submissions and powers the authenticated admin tools we use to read them. - Cal.com: runs the booking calendar embedded on the contact page and sends us booking events. - Brevo: sends the confirmation and notification emails. - PostHog: product analytics for the marketing site, so we can see which pages are used and where the flow breaks. - Google Analytics: aggregated traffic measurement and page-level usage of the marketing site. - Vercel: hosts the site and keeps request logs. ## How long we keep it Contact enquiries and booking details are kept for as long as needed to respond and for reasonable record-keeping, and are deleted on request. Hosting logs are retained on the provider’s default schedule. ## Your rights If you are in the UK, the EEA, or a jurisdiction with similar law, you can ask us to: - Access: send you a copy of the personal data we hold about you. - Rectify: correct anything inaccurate or incomplete. - Erase: delete your data where we no longer need it. - Restrict or object: limit or stop processing based on legitimate interests. - Port: give you your data in a structured, commonly used format that a computer can read. ## Security The site is served over encrypted connections (TLS). Access to submissions is limited to named Falnor staff behind authentication. Client engagement data is never hosted by Falnor; it stays in the client’s own systems under the client’s controls. ## Cookies and Do Not Track The site sets cookies and local storage for PostHog product analytics and Google Analytics, which measure how the marketing site is used. The embedded Cal.com booker may set functional cookies needed for the booking to work. We do not run advertising or cross-site tracking. We do not change site behaviour in response to a Do Not Track signal. ## Minors and changes This site is not intended for anyone under 18. We may update this policy when our practices or the law change; material changes will be noted on this page with a new date. ## Contact To exercise any of the rights above, or to ask about this policy, email support@falnor.io. We respond to verified requests within 30 days. # Terms and conditions (https://falnor.io/terms) Updated September 2026. falnor.io is owned and operated by Falnor Technologies LLC. Using this site, or engaging Falnor for services, means you accept these terms. Specific engagement terms are set out in the statement of work you sign with Falnor, which takes precedence over these site terms if they conflict. ## 1. What Falnor does Falnor designs and builds systems, one workflow at a time, inside the software a client already uses. We deliver a working system, set up on the client’s own systems and tested on the client’s real cases before it takes live action. We do not sell software licences, seats or subscriptions, and we do not resell third-party products. ## 2. Engagements and scope Each engagement covers one named workflow and is scoped after a discovery call. Scope, deliverables, success criteria, go-live gates and price are written into a statement of work agreed before any work begins. Work outside that statement of work is quoted separately. ## 3. Fees and payment Fees are charged for the system we build, and agreed in writing after mapping, before any build starts. There is no usage meter and no rate card. Payment is made in stages against the milestones in the statement of work. AI and hosting costs incurred by a live system run through the client’s own accounts at provider list price, with no markup. We do not collect payment card details on this site. ## 4. Ownership and the exit pack The client owns the deliverables produced under an engagement: the links to the client’s software (connectors), the workflow definitions, the standing rules, the full record of actions (evidence log) and the test cases (regression set), delivered on the client’s own systems in plain formats. Handing these over (the exit pack) is part of the scope. The system keeps running in the client’s systems under the client’s accounts and does not depend on a Falnor-operated service. Falnor keeps ownership of its pre-existing tools, methods and intellectual property developed independently of the engagement. Nothing here licenses the resale or redistribution of Falnor’s own tooling. ## 5. Warranty and liability Beyond the express success criteria, re-testing and support commitments in a statement of work, services and site content are provided as is, to the extent the law allows. We do not guarantee compatibility with every third-party system; compatibility is checked during discovery before an engagement is agreed. To the extent permitted by law, Falnor is not liable for indirect, incidental or consequential loss, including loss of data, profit or business interruption, arising from use of this site. Liability under an engagement is capped and allocated as set out in the statement of work. ## 6. Suspension and termination Falnor may suspend or end access to the site or an engagement for: - Abusive, defamatory or malicious behaviour towards Falnor staff or clients. - Unauthorised resale, distribution or promotion of Falnor’s proprietary materials. - Hacking, spam, piracy or other unlawful activity. - A material breach of the statement of work or these terms that is not cured within the notice period. ## 7. Privacy How we handle personal data collected on this site is described in the privacy policy. Client engagement data processed in the client’s environment is governed by the statement of work and security review for that engagement. ## 8. Governing law and contact These site terms are governed by the law stated in your statement of work or, where there is none, by the law of the jurisdiction in which Falnor Technologies LLC is registered. Questions about these terms: support@falnor.io.