Set up the Claude Project that knows your agency, so every prompt after this one answers as you rather than generically. Before anything else, you give Claude enough about you to be useful. In this lesson you create a Claude Project and load it across three pillars — you, your offer, and your niche — using an interview prompt that asks you twenty questions and writes the pillars for you. It takes about ten minutes and it is the reason every prompt in the rest of the course produces something specific to your agency instead of something generic. Download: the Agency Brain starter pack, three pre-headed pillar templates ready to paste in.
Step by step walkthrough using Claude.
Twelve prompts you will use for the rest of this course — and the rest of the agency. A prompt you rewrite every time is not an asset. Here you build a library of twelve prompts organised under four headings — Decide, Write, Build and Sell — covering niche scoring, pricing, cold outreach, bot scripts, automation planning, objection handling and retainer pitches. You also run the voice-rules prompt so everything Claude writes sounds like you, and adapt your first three prompts to your own niche. Download: the Foundation Prompt Library, all twelve prompts in one copy-paste file.
Every prospect asks a version of the same question: why should we pay you for something we could probably work out ourselves? This lesson gives you the answer. You'll learn the three forces driving demand — adoption is climbing, capability hasn't caught up, and almost nobody is providing implementation — then turn all three into language a sceptical buyer actually responds to. It closes with the objection you'll hear most: that they'll just use Claude themselves. Download: the Market Case one-pager, to send after a discovery call.
Most people picture this business wrong. You are not building AI from scratch — you're importing a proven template, customising it for one business, handing it over properly, and charging monthly to keep it running. This lesson walks through that cycle end to end, so you know what a working week actually looks like and where the money comes from at each stage. Download: the service catalogue — five chatbot types and ten automations, each with a realistic price band.
You can run this business alone and earn well. You can also outsource and grow faster while earning less per job. Neither is wrong, but choosing by accident is. This lesson works through what your hour is genuinely worth, which tasks to hand off first — and which never to hand off — where to find people who can do the work, and what it costs. Download: the outsourcing decision grid and a freelancer brief template.
Selling AI to everyone means saying nothing specific to anyone, which is why the messages get ignored. This lesson covers what actually makes a niche buyable — enough businesses to sell to, enough repetitive admin to be worth automating, an owner you can reach, and a budget that already exists — then works through seven candidates including real estate and e-commerce so you can see the criteria applied rather than just described. Download: the niche shortlist worksheet.
Most people pick a niche, doubt it a week later, and start again. This lesson ends that loop. You'll score each candidate against five criteria for a mark out of twenty-five, and commit to the winner on paper with the reasoning written down — so when the doubt arrives, you reread the score instead of restarting. Twenty minutes, and the decision is made. Download: the Niche Scorecard, with a fully scored worked example.
The gap between knowing your niche and having something to sell them is where most agencies stall. Here you name one service, scope exactly what's included, decide what's deliberately excluded, and set a build price and a monthly retainer you can say out loud without hedging. You'll also get the answer for when a prospect asks for something outside it — which happens on almost every early call. Download: the offer one-pager template, with a completed example.
Support, sales, lead generation, e-commerce and booking. This lesson covers what each one does, which businesses actually pay for it, and roughly what it sells for — so when a prospect describes their problem you already know which bot you're quoting. It also settles which type to build first, because the easiest one to sell and the easiest one to deliver aren't always the same. Download: the chatbot type matrix, mapping niche to bot type to price band.
The reason this business works: the hard build is already done and sitting in a template library. Here you'll watch a template imported into VoiceFlow start to finish, then do it yourself. You'll also learn the checks to run before you import — what the template already does, what it doesn't, and what you'd have to change — because the wrong starting point costs more time than building from scratch. You finish this lesson with a working bot in your account. Download: the template shortlist, vetted by bot type.
An imported template is not a deliverable. In this lesson you turn one into a real bot for a real business: Maya's boutique gym. Branding, conversation copy in her voice, the logic behind class bookings, and the ten questions her members actually ask going into the flow. This is the full build, start to finish, on a client you'll follow through the rest of the course. Download: the Pebble & Co brief and the finished flow, ready to import.
The build is the easy part. This lesson covers what separates a paid job from an unpaid one: the test script you run before anything goes live, what you actually give the client at handover, and how you set up the monthly retainer in the same conversation. Handover done well is how the £250 a month starts — done badly, it's where the client decides they can manage it themselves. Download: the bot test script and the client handover document template.
This is why running AI Agenices is easy. Here I show you how to import an AI Agent template into Voice Flow.
Two platforms that do broadly the same job, and one decision you only need to make once. This lesson covers what each is genuinely good at, where the free tiers run out, and when the open-source option is worth the extra setup — versus when it's a distraction that costs you a week. Pick one, learn it properly, and move on. Download: the platform comparison card.
CRM sync, lead capture, auto-responders, meeting scheduling, internal alerts, document generation, lead scoring, onboarding flows and more. For each one you get what it does, which businesses need it, and — the part that matters — the single sentence that sells it to an owner who doesn't know they want it yet. Download: the Ten Automations reference sheet, each with a price band.
An end-to-end build, done with you. Maya loses enquiries every night after nine because nobody's there to answer them. In this lesson you build the automation that fixes it: the enquiry is captured, logged to her CRM, answered with a reply that sounds like her, and flagged to her team the next morning. You finish with a live automation running, not a diagram of one. Download: the importable Make.com blueprint and the build-along steps.
Everything you build will fail eventually — a token expires, an app changes, a field goes missing. This lesson is about what happens next. You'll learn what to test before handover, how to build error handling in so failures surface instead of going silent, and how monitoring becomes a line item on the retainer rather than free work you do at weekends. Download: the pre-handover test checklist.
You need less than you think. This lesson lists what you genuinely need to take on a first client — the build platform, the AI, somewhere to store client work, and a way to get paid — with realistic monthly costs against each. Everything else is a decision for later, when a client is paying for it. You'll also see which free tiers hold up under real client work and which quietly don't. Download: the stack checklist with indicative costs.
The template you start from decides how much of the job is left. This lesson covers where to find them — the marketplaces, the communities, the vendor libraries — and the four checks to run before you commit: what it already does, what it doesn't, how much rework you're inheriting, and whether it can do the one thing the client actually asked for. Download: the template evaluation checklist.
The unglamorous lesson that saves you a client. Who owns the VoiceFlow account? Whose card is on the Make.com subscription? What happens if they leave? This lesson sets out what sits in your account, what sits in theirs, and how to ask for access without sounding like you're asking for the keys to the business — because that arrangement quietly decides whether a retainer is defensible or whether they can walk away with everything next month. Download: the client access request template and an ownership policy you can adopt as-is.
The lesson that stops you undercharging. Four ways to price this work — fixed, hourly, value-based and retainer — with a straight answer on when each one fits and which to lead with. More usefully, it gives you the actual numbers: what a bot build goes for, what an automation goes for, and what changes the figure. You finish knowing what your next quote says. Download: the price band reference card.
A price is only high or low next to something. This lesson teaches you to build that something out of the client's own numbers: hours saved a week, errors removed, enquiries recovered that used to go unanswered. Work it through on Maya's gym and the £1,200 stops sounding like a cost and starts sounding like arithmetic. You'll leave with the three figures you say out loud on a call. Download: the ROI calculator with the worked example.
One-off builds pay once. The retainer is what turns this into a business you can plan around. This lesson sets out three tiers, exactly what goes into each so it never becomes open-ended, and how to raise it in conversation without it landing as an upsell. Crucially, it covers when to raise it — because that conversation is far easier before handover than three months after. Download: the three-tier retainer menu template.
The mistake that costs new agencies weeks: building the whole thing hoping it wins the deal. This lesson shows you how much to build to close — a light demo of one specific pain, not a finished product — and how to confirm scope in writing before you do the real work. Nothing deep gets built before a signature. Download: the demo script and the scope-confirmation email.
Four routes to a first client: inbound, outbound, local, and the marketplaces. This lesson covers what each one actually costs you in hours, how long before it produces a reply, and which suits the time you genuinely have this month. Most people pick the slowest channel by accident, then conclude the business doesn't work. Download: the lead source map, with effort and lead time for each channel.
Three channels, one method. Fix the profile so it says what you do, build a list of businesses that fit your niche, send a message that mentions their actual problem, and follow up properly instead of once. This lesson runs those four steps across all three channels, so you're not learning three systems — you're learning one and applying it three ways. Download: the outreach swipe file, nine messages across the three channels.
Thirty minutes that decide whether there's a job here. This lesson gives you the twelve questions to ask, in order, so the client describes their own problem in enough detail to scope it. It also covers when to show the demo, how to talk about price without committing early, and how to close the call with a specific next step rather than "I'll send something over." Download: the discovery call question sheet and note template.
Long proposals don't get read and they don't get signed. One page: the problem in their words, what you'll build, what it costs, how long it takes, and what happens next. This lesson walks through a completed example — Apex AI to Pebble & Co — and covers the thing that matters more than the format, which is sending it the same day while the call is still fresh. Download: the one-page proposal template, filled in.
The plan that turns everything in this course into a client. Days one and two: niche scored, offer written, price decided. Days three to five: demo bot built and live. Days six and seven: list built, first outreach sent. Then the Momentum Plan takes you through days eight to thirty — calls booked, proposal sent, first client, first retainer conversation. Nothing here is promised faster than it can honestly be done. Download: the 7-Day Sprint tracker and the Days 8–30 Momentum Plan.