# Enabled by AI: the full site in one file > Enabled by AI helps small and mid-sized companies stay competitive by augmenting the people they already have with agents and AI workflows, built into the software they already own. Every engagement starts with a free AI Opportunity Audit: forty-five minutes that return the three jobs most worth automating, ranked. Where the numbers justify it, it continues into a one-day AI Enablement Workshop with the leadership team. It is an AI consultancy based in Cardiff, working with companies across South Wales first and delivering UK-wide. Generated 2026-08-15 from the same data the pages render. The curated index is at https://enabledbyai.co.uk/llms.txt. ## The field note the site is built on Agents do not replace your team. They delete the work that was never worth your team's time. ## The offer, in figures - 45 minutes: To see which of your workflows an agent should take. - One day: From ranked audit to agents running in your own software. - 30 days: Of tuning while your team runs them on live work. - 3 modules: Work Audit, Use Case Design, Installation. ## Why this is urgent Your competitors got faster. Same headcount. It does not announce itself. A firm your size quietly stops losing Friday to the weekly numbers. Quotes get followed up the same day, every day. Invoices get chased on time whether or not anyone had a busy fortnight. None of that came from better people. It came from work that no longer needs a person: the chasing, the rekeying, the first draft, the report nobody enjoys assembling. Their team now spends its hours on the half of the job that actually needs judgement. You are competing with that. Every week run the old way is a week of compounding difference, and it is far easier to close a two-month gap than a two-year one. ## The symptoms - You bought licences for everybody, and you could count the habitual users on one hand. - Three departments are quietly running their own experiments, and nobody can say which are worth keeping. - The same figures are rekeyed between two systems every week, and everyone stopped mentioning it years ago. - Your best people lose the first hour of every day to work that carries no judgement in it. - A client, an insurer or an auditor asked what your AI policy is, and the answer took a fortnight to write. - You have sat through a demonstration that was genuinely impressive, and still could not say what it would change on Monday morning. ## The free AI Opportunity Audit Duration: 45 minutes. Price: No charge. Preparation: Six questions, about ten minutes. 1. Six questions before the call. What the business does, which systems it runs on, and where the week goes. It takes about ten minutes and it is the only homework there is. 2. Forty-five minutes with Peter. He walks you through the recurring work he can already see in your answers, and the three jobs he would automate first, with the hours each one costs you now and an honest read on how much of it a machine can take. 3. The ranking, in writing. You keep it whether or not you ever book anything. It is a first pass over the same audit that opens the workshop, so you are seeing a sample of the actual deliverable rather than a brochure. 4. Then a straight answer. If a workshop would pay for itself, Peter says so and quotes it on the call. If your best next move is something you can do yourselves in an afternoon, he says that instead, and you are welcome to go and do it. ## The AI Enablement Workshop: three modules - Work Audit (Before the day): The full version of the audit you have already seen a first pass of, across the whole team: every recurring job, scored and ranked by how much of it a machine can safely take. - Use Case Design (The morning): Your leadership team decides which two or three jobs go first, and we design each one end to end: what starts it, what the AI drafts, what a person signs off. - Installation (The afternoon): The chosen jobs are built into the software you already pay for and run end to end on real work while you watch. ## The method: what every recurring job is scored against - Frequency: What repeats: daily, weekly, every invoice, every enquiry. Rare work is rarely worth automating, however annoying it is. - Hours: How long it takes and how often it runs, multiplied out across the year and attached to the person actually doing it. - Handoffs: Where work stalls, duplicates or quietly disappears between two people or two systems. This is where most of the loss hides. - Judgement: How much of the job is following a rule and how much is deciding something. The split determines what a machine may touch. - Blast radius: What it costs if it goes wrong in front of a customer. High blast radius does not mean leave it alone. It means a person approves it. ## The three autonomy modes Every workflow is assigned one of these, in writing, on the day: - The machine runs it: Repetitive, high volume, no judgement, and cheap to get wrong. Sorting, extracting, rekeying, chasing. Nobody needs to read the output of these, and nobody wants to. - The machine drafts, a person ships: Anything a customer will read. The AI does ninety per cent of the work and one of your people presses send, because your reputation took years to build and a poorly judged message undoes it in an afternoon. - A person keeps it: Judgement, negotiation, bad news, anything where being wrong is expensive. We deliberately leave these alone, and saying so out loud is what makes the rest credible. ## How the first workflows are put into service 1. Name the person who owns each job and has the authority to change how it is done. 2. Give the system the access, the examples and the tone it needs, taken from your own work rather than a template. 3. Run it alongside the current process first, so its output is compared against reality before it acts on anything. 4. Pick the one number that says whether it is working, and check it weekly until the team stops checking. ## The pattern behind it - Before: A recurring pain gets a job description, because that is the instrument every company has to hand. - Audit: Write down the workflows that role would actually spend its week on, and what each one costs today. - Build: Design against that list, and decide for each one whether a machine runs it, drafts it, or keeps out of it. - After: The repeatable work runs itself and reaches a person prepared. The judgement stays where accountability already sits. ## Where we usually start: the recurring use cases ### Finance: The chasing nobody enjoys Today: Somebody exports the aged debtors report, works out who to chase, writes broadly the same email eight times, and loses track of who replied. It happens late, because it is nobody's favourite Tuesday. What gets built: Overdue invoices identified from your accounts system, chased on a schedule in your company's own tone, escalated to a person the moment a reply needs judgement, and reconciled back so the ledger stays the source of truth. What changes: The chasing happens on the day it should, whether or not anyone had a busy fortnight. The judgement calls still reach a human. There are simply far fewer of them. Autonomy mode: The machine drafts, a person ships. ### Revenue: The follow-up that never happens Today: A quote goes out. The intention is to follow up on Thursday. Thursday is busy. Three weeks later somebody notices, and the message that finally goes is apologetic rather than useful. What gets built: Every quote and enquiry tracked through to a reply, with the follow-up drafted and the context already gathered: what they asked for, what was quoted, what has happened since. Nothing is sent until whoever owns the relationship has read it. What changes: Nothing goes cold through inattention. The person who owns the relationship spends their time deciding what to say rather than remembering to say something. Autonomy mode: The machine drafts, a person ships. ### Leadership: The Monday brief Today: Two people spend Friday afternoon assembling numbers from four systems into a deck that is read once. By the time the decision arrives, the figures are a fortnight old. What gets built: The week's numbers pulled from the systems they already live in, assembled into the same short brief every Monday morning, with the exceptions surfaced rather than buried and the source figures traceable. What changes: The reporting stops consuming a person, and the meeting starts from what changed rather than from what the numbers are. Autonomy mode: The machine runs it. ## How we work - We augment the team you have. We do not replace it. Every workflow we build gives a named person more capacity, not less standing. The machine takes the repetition and hands the work back prepared. Your people keep the customer, the judgement and the credit. - Designed for your business, not taken off a shelf. There is no product here to sell you. We look at the work your company actually repeats every week and design around that, which is why the first thing that happens is an audit rather than a demonstration. - You know what is coming, and what it costs, before anything is built. The scope is agreed and the fee is fixed before a single prompt is written. No day rates that creep, no discovery phase that quietly becomes the project, and no invoice that arrives larger than the conversation that preceded it. - We do the technical work. Your team is trained on the result. You do not need a developer, and there is no code left behind for anyone to maintain. The people who will run each job are in the room while it is built and run it themselves before the day ends, trained on their own work rather than thrown in cold. - It lives where your team already works. Inside Teams, Slack, your inbox, your CRM: whatever is already open on their screens. Nobody logs into a new dashboard, because the systems that get used are the ones that turn up where the work already is, behaving like another member of the team. - It gets better the more it is used. The first version is corrected against real work in front of you, then tuned again over the thirty days that follow as your team runs it on live jobs. What starts as a decent draft becomes something that sounds like your company. ## Who runs it - He installed this in his own company first. Peter O'Donoghue runs an AI company whose research, reporting, chasing and weekly numbers are carried out by a fleet of AI agents he built and supervises. Nothing you are shown is theoretical. It is the way his own business runs. - He knows where AI stops being useful. Twenty years inside B2B firms before he founded an AI product, which means he knows precisely where the technology earns its keep and where it will embarrass you in front of a customer. The work comes first and the AI comes second, in that order. - He runs it himself, start to finish. There is no delivery team behind the curtain. Peter does the audit, runs the day and answers every question in the month that follows, which is the reason only a handful of companies are ever underway at once. ## Deliverables - A ranked map of the work AI can take on. Every recurring job in the business set out by name, with the hours it consumes, who does it, which systems it touches, and an honest read on how much of it a machine can take without a person losing control of the outcome. Delivered before the day, not during it. - Two or three jobs running inside your own software. Built in the afternoon, in the tools your company already pays for: your email, your documents, your chat, your CRM, your finance system. There is no new subscription and no new interface for anyone to learn. Each one runs end to end on real work before anybody goes home. - A written playbook and an AI policy your company keeps outright. Every job set out in plain English: what starts it, what the AI does, what a person checks, and what must never go near a model. Enough to hand a new starter, and enough to show a client or an auditor who asks. There is no licence and nothing to renew. - A team trained on what was built, rather than left to work it out. The people who will run each job are in the room while it is built, and they run it themselves before the day ends. Training happens on your own work rather than on a demonstration, which is the difference between a team that uses a system and a team that quietly goes back to doing it by hand. ## The process, start to finish 1. The AI Opportunity Audit, at no charge. Six questions, then forty-five minutes with Peter. You leave with the three jobs worth automating first, ranked, in writing, and a straight answer about whether a workshop is worth your money. Most of what follows only happens if that answer is yes. 2. Scope and paperwork. The fee is fixed on that call against a fixed scope, and confirmed in a letter of engagement with an invoice. It is payable in full before the audit work begins. 3. Intake and access. You complete a short intake form covering what you sell, who does what, and how the week runs. A read-only look at the tools you already pay for tells us what is already possible without buying anything. 4. Work Audit. Before the day, Peter maps the recurring work across the whole team and returns it ranked three ways: by the hours it consumes, by how much of it a machine can safely take, and by what it would cost you if it went wrong. 5. Use Case Design. On the morning of the workshop your leadership team chooses which jobs go first: two or three rather than ten. We then agree what triggers each one, what the AI is allowed to draft, what a person must always approve before anything reaches a customer, and which information never leaves the building. 6. Installation. In the afternoon the chosen jobs are built into the tools you already use, in the room and on screen. Real work from your own week goes through them end to end while you watch, and you correct the output in front of us. We then set the weekly hour, name who owns each job, and hand over the playbook, the prompts and the policy. 7. Implementation. Your team runs the system on live work for thirty days. Peter corrects the prompts and the rules himself as the first output comes through, answers questions by text or email within a working day, and closes with a review call on day thirty. ## The day, hour by hour Times are indicative; the order is the commitment. - 09:30 What the audit found: We open with your own work in front of you: every recurring job, the hours it takes and how much of it a machine can take. Most teams spend the first twenty minutes arguing with the numbers, which is exactly the right use of them. - 10:30 Choosing what goes first: Two or three jobs, chosen by the people who own them. We are looking for work that is repetitive, high volume and low judgement, while deliberately avoiding anything where being wrong is expensive. - 11:30 Designing the guardrails: What the AI is allowed to draft, what a person must approve before it reaches a customer, and what information never leaves the building. This is the half hour that keeps you out of trouble, and it is written down. - 13:30 Building it: The jobs are built in your own tools, on screen, in the room. No sandbox, no demo environment, no slide of what it could look like. - 15:00 Running it on real work: Real work from your own week goes through end to end while you watch. You correct the output in front of us, and we tune it until your team recognises it as their own standard. - 16:15 Handover: Who owns each job, what the weekly hour looks like, and the playbook, the prompts and the policy handed over. You leave owning it. ## Logistics - Where to start. The AI Opportunity Audit: forty-five minutes, no charge, and you keep the ranking whether or not you go further. - Duration. One day in the room, plus thirty days of implementation support. - Scheduling. A workshop can usually be underway within two to six weeks, depending on the time of year. - Where. At your offices, anywhere in the UK, with travel included in the fee. Online if your team is split across sites, in the same hours. - Participation requirements. One director completes the intake form, roughly thirty minutes. Three to six people attend the full day, 09:30 to 17:00. Bring the people who own the work, not only the people who own the budget. The team gives roughly one hour a week each across the thirty days that follow. - What you need in place. Administrator access to the software you already use, so the jobs can be built rather than described. Nothing else. We do not ask you to buy anything to take part. - Fee. One fixed fee for the whole engagement, quoted on the audit call once the scope is clear. No tiers, no day rates, no licence and nothing to renew. - Payment terms. Prepaid and non-refundable. Payable in full by bank transfer before the audit work begins. ## The second service: the Agent-Ready Install Your buyers already ask ChatGPT, Claude and Perplexity who to use before they open Google, and the companies you want as customers are starting to send procurement agents to do the choosing for them. If those assistants cannot find you, trust you and quote you, the enquiry goes to a competitor and you never hear about it. The Agent-Ready Install makes sure the recommendation is you. Your next enquiry will be sent by an agent. The way companies choose who to work with is changing hands. Today a person asks an assistant for a shortlist and phones the top of it. Next the person does not do the research at all: their procurement agent reads the market, compares the candidates, rules out anyone it cannot verify and comes back with a recommendation and the reasons. An agent doing that job does not browse. It does not admire your design, forgive a vague page or call to clarify. It reads what machines can read, checks what machines can check, and moves on. A business that is impressive to people and invisible to agents loses deals it never knew existed. One that is easy to find, trust and quote gets shortlisted while its competitors wonder where the enquiries went. What an assistant needs before it recommends anybody: - Found: An assistant recommends from what it can find, and most company websites are invisible to the tools buyers now ask, for reasons nobody inside the company can see. We make sure the questions your buyers ask have answers with your name on them, in the places assistants actually look. - Trusted: An assistant will not stake its recommendation on a business it cannot check. We put your records straight everywhere machines verify you, so that looking you up confirms you and recommending you feels safe. - Quoted: An assistant has seconds on your site to find something worth repeating. We turn what you already know into plain answers it can lift word for word, so the version of you it passes on is the one you would have given yourself. The engagement: 1. The visibility audit. We put your buyers' questions to the major assistants and show you the answers unedited: where you do not appear, what they get wrong about your prices and your fit, and which competitor is handed your enquiry instead. Most owners have never seen this. It is usually the whole conversation. 2. The install. In about a day, your business becomes easy for every major assistant and agent to find, trust and quote. Nothing about how your site looks to a person changes unless we agree it should, you do not need a developer, and there is nothing for your team to learn. 3. The measurement loop. Every month the same questions go back to the same assistants, and you see the movement: where you now appear, what they now get right, and what to fix next. Keep it running or take it in-house. Either way you can answer the question your board will ask: when buyers ask AI, do we come up? What gets installed: - Your site becomes readable to every AI assistant, not only Google. Most modern websites are blank pages to them; yours will not be. - The files agents look for first, written for you and regenerated automatically, so what AI says about you cannot drift from the truth. - Your services, prices and answers in a form assistants repeat accurately, so you stop being misquoted or skipped. - Your business profiles and listings put straight, so an agent checking up on you finds you confirmed. - Your site registered with the search indexes that feed the assistants, including the one behind ChatGPT that almost everyone forgets. - One press angle worth pitching, because an independent mention is worth ten pages of your own. - A written before and after: what the assistants said about you on day one, and what they say at day thirty. This site is the demo. Ask any assistant to read enabledbyai.co.uk/llms-full.txt and describe what we do. That file regenerates from the site's own data on every deploy, and everything offered on this page was installed here first. - Where to start. The AI Opportunity Audit: the same forty-five minutes tells you whether the bigger bleed is inside the business or in front of it. - Duration. The audit and the install together usually take a day of our time and very little of yours. The measurement loop is monthly, for as long as it earns its keep. - What we need. Access to the website and wherever it is hosted. An hour with whoever owns the company's profiles. - What you keep. Everything: the files, the records, the checklist and the before and after. No licence, nothing to renew. - Fee. Fixed, quoted on the audit call once the scope is clear. ## Frequently asked questions ### The audit Q: What is the catch with a free AI Opportunity Audit? A: There is no catch, but there is a reason. The audit is the part of this work that sells the rest of it: once a leadership team can see the hours their recurring work actually costs, ranked, with their own job titles attached, the decision usually makes itself. Peter would rather spend forty-five minutes finding out whether that is true for your company than spend it persuading you of anything. If it turns out there is not enough there, he will say so, and you keep the ranking either way. Q: What does the audit actually cover? A: Six questions beforehand, covering what the business does, which systems it runs on and where the week goes, then forty-five minutes on the recurring work those answers reveal. You leave with the three jobs worth automating first, the hours each one currently costs you, an honest read on how much of it a machine can safely take, and what would have to be true for it to work. It is a first pass over the same audit that opens the workshop, so you are seeing a sample of the real deliverable rather than a brochure. Q: What is the wait time? A: An audit call is usually available within a week. Beyond that, Peter runs every workshop personally and supports every company through the month that follows, so only a small number are ever underway at once. In practice that means two to six weeks before a date becomes available. If you have a planning cycle or a budget year approaching, it is worth having the call earlier rather than later. ### The day Q: Can a single day really be enough? A: The day works because it does not start from nothing. The audit that would otherwise fill the first fortnight is finished before you arrive, so every hour in the room is spent deciding and building rather than gathering. You are also not being taught a framework to apply later. The jobs are designed, built and run on real work from your own week before anybody goes home. The thirty days afterwards exist for the one part that genuinely cannot be compressed, which is your team using the system for the first time on live work. Q: How much of my team's time does this take? A: Forty-five minutes to find out whether it is worth doing at all, and after that one working day plus about half an hour of intake. The workshop runs from roughly half past nine to five, with a break for lunch. Then the commitment is about an hour a week each, and Peter carries the tuning work through the following month so it does not land on anyone's desk. Q: Who should be in the room? A: Three to six people: whoever runs the business, and whoever actually does the work we are going to automate. The second group matters more than the first. A day spent with directors alone produces a system nobody uses, because the person who does the job every day is the only one who knows the exceptions that break it. ### Your team and your data Q: Will this cost people their jobs? A: Not in any company we would take on, and it is a fair question to ask before you put it to your team. We deliberately choose work that is repetitive and carries no judgement: rekeying, chasing, first drafts, looking things up. That is where a machine is genuinely better, and where nobody enjoys spending their afternoon. If your plan is to cut headcount, you do not need a workshop for that and we are the wrong people to run it. Say plainly to your team what the day is for. The ones who do the job usually arrive with the best list of what to automate first. Q: What happens to our data? A: You decide, in writing, on the morning of the workshop, and it goes into the policy you keep. In practice we build inside the tools you already trust and already have a contract with, using the business tiers where your content is not used to train anybody's model. Anything genuinely sensitive, such as client files, personal data and the things your insurer would ask about, is either kept out of the system entirely or handled by a person. That decision is yours, we write it down, and you can hand the result to a client or an auditor who asks. The free audit needs none of it: six questions about how the week runs, and no access to anything. Q: Do we need a technical person, or developers? A: No. We work with the people who run the business and the people who do the work, not with developers, and there is no code for anybody to maintain afterwards. Peter does the technical part on the day. Your team needs to know how to use what has been built, which is what the afternoon and the following thirty days are for. If you do have someone technical, bring them. They usually enjoy it and they will own it well, but nothing here depends on having one. Q: Do we need to buy new software? A: No. We build inside the tools you already use, because software you have already paid for and already know how to open is far more likely to be used. Most companies find that what they need is sitting inside a subscription they are already paying for. We only recommend anything new when a genuine gap is blocking the work, and that is rare. ### The fee and what follows Q: How much does the workshop cost? A: One fixed fee for the whole engagement, quoted on the audit call. It is fixed because the scope is fixed: the full audit before the day, the day itself, the jobs installed inside your own software, the playbook and the policy your company keeps, and thirty days of support while your team runs it. There are no tiers, no day rates that creep upwards and nothing to renew. The reason it is quoted rather than published is that the honest answer to whether it is worth it depends on what the audit finds, and neither of us knows that before the call. Q: Why is the fee prepaid and non-refundable? A: Because the largest single piece of work happens before you walk into the room. The full audit of your recurring work, the read of your systems and the first ranked list of what to automate are all finished in advance, and that work cannot be unspent once it is done. Paying in full before it begins keeps the scope fixed and the fee fixed with it. Nobody is asked to commit to any of that until the free audit has shown them what is there. Q: What does the support in the thirty days actually cover? A: Three things. Peter reviews the output as your first real work goes through and corrects the prompts and rules himself, so the tuning does not become somebody's homework. He answers questions by text or email within a working day. Then at the end of the month there is a review call, where we look at what the system produced, adjust anything that is not working, and confirm the weekly hour is holding without him. After that there is nothing further to pay and no licence to renew. Q: What happens after the thirty days? A: Nothing, unless you want it to. The capability ends up with your company rather than with us: the playbook, the prompts and the policy are written in plain language and kept inside your own tools, so nothing is locked in an account you cannot reach and nobody has to call us to change a prompt. Some companies do ask Peter to keep going month by month afterwards, adding jobs as the first ones bed in. That is a conversation you have once you have seen the work, and it is never a condition of booking. ### What this is not Q: Is this an AI automation agency? A: No. The work comes first and AI provides the leverage. The AI handles what is repetitive, high volume and low judgement. A person keeps the judgement and approves anything that reaches a customer, because your reputation took years to build and a poorly judged message can undo it in an afternoon. If what you want is a bot answering your customers unsupervised, we are the wrong firm. Q: Could we not just do this ourselves with ChatGPT? A: You could, and a good number of companies are trying. What usually happens is that two or three enthusiasts build something impressive, it works while they maintain it, and it quietly stops the month they get busy. The difficult part was never the prompt. It is choosing the right jobs, agreeing what a person must always approve, building it where the work already happens rather than in somebody's private account, and writing it down so the company owns it rather than an individual. That is what the day is for. If the audit shows you can genuinely do it yourselves, Peter will tell you so. ## Where An AI consultancy based in Cardiff, built for South Wales companies first: Cardiff, Newport, Swansea, Bridgend and the Valleys. Workshops run at your offices anywhere in the UK with travel included in the fee, or online where a team is split across sites. ## Pages - Home: https://enabledbyai.co.uk/ - The method: https://enabledbyai.co.uk/method - Use cases: https://enabledbyai.co.uk/use-cases - The workshop, scope and terms: https://enabledbyai.co.uk/workshop - The Agent-Ready Install: https://enabledbyai.co.uk/agent-ready - FAQ: https://enabledbyai.co.uk/faq - Book the free audit: https://enabledbyai.co.uk/consultation ## Contact Peter O'Donoghue. Text +44 (0) 7967 974777. Lambourne House, Lambourne Crescent, Cardiff CF14 5GL, United Kingdom. The form at https://enabledbyai.co.uk/consultation emails Peter directly and stores nothing.