How AI can accelerate our engagements, amplify our marketing programmes and optimise our operations. Three tools that run the studio. Six programmes we sell.
The AI we run on ourselves first — the reason the work goes faster and the margin holds. Each is a brain: one for the numbers, one for the judgment, one for the people.
The operating brain. Instead of CRM, time tracking and accounting sitting as separate islands, Br(ai)n connects them into one causal model: who's on what, what it's consuming, where margin is being made or quietly burned. The firm sees the economics of every engagement in flight — not three months later when the accountant closes the month. Saying "yes", "no" or "not on those terms" to new work becomes a number, not a gut call.
Concept →The house mind. Grounded on our own archive — old campaigns, pitches, decks, brand work, client correspondence and the reasoning behind them. Recall what we pitched a client and why it won. Mine the archive for analogous work when a brief lands. Draft in our voice. Let a new hire query James instead of interrupting the founders. It turns the knowledge locked in heads and folders into a compounding asset — and means the firm stops being one person.
Concept →The bench. The third brain — who can we get to deliver this. A live database of UK marketing specialists you ask in plain English ("senior freelance copywriter, London, drinks & FMCG"), with location and LinkedIn revealed on the ones worth a call. It plugs into Br(ai)n: when an engagement needs a skill the team doesn't have free, the bench answers it. Built as a working proof over a 2,000-strong pool.
Open the bench →Each solves one known problem and starts small — cheap, low-risk, measurable in weeks. A practitioner stays in the room; AI does the grunt work behind them. Tagged Built (ready to drop in) or Co-build (the shared-IP conversation).
The land offer. Twenty days: on-site extraction, AI analysis, a board-ready insight hub. We run the audit lenses that matter to an owner heading for exit — margin, pipeline, CAC and lifetime value, dormant value, AI visibility. The data Vault and brain it builds aren't binned at handover; they're what the retained programme is built on. The audit is the first half of the work, not a separate cost.
See the motion →The board-room opener. Type in the budget; the model runs the channels and shows the growth curve and what it does to the company's valuation. It plans the year and it sells the next engagement — for an owner heading for an exit, it speaks the only number that matters. After the audit it becomes the live report that shows the return every month.
Detail →The big-catalogue opener. AI takes messy sales, cost and product data and hands back a clean margin map — by product, customer and channel. Read-only, days not weeks, measured in pounds of margin found and dead lines to cut. The strongest way into a large-catalogue client.
Detail →The margin lever. Take one job a professional-services firm is buried under — proposals, pitch production — and treble the output with AI on top of proven method. Tightly scoped. We run it on ourselves first (that's how the practice holds its margin), then sell the proven thing.
Detail →The fastest proof. Wake up the dead customer base: AI sorts the list, picks the ones worth a call and drafts the opener — their own people send it. No mass blast, because that's the part senior buyers bin. Quick cash, low risk, measured in pounds brought back. Proven on a live client with 2,000 dormant contacts — the fastest way to show AI earns its keep.
Open Gold Digger →The forward bet. Factual, sourced content on domains the firm owns, written to get cited inside AI answers — where senior buyers now do their first research. Cheap to run, measured in citations and share of answer. Nobody in the UK mid-market has claimed it yet.
Detail →Agencies sell people and hours. We sell a number, then prove it — and AI is what keeps the retainer profitable.
The three brains and the six programmes don’t sit on nothing — they run on one substrate: Atrapos, Fosferon’s judgment layer. A governed layer above the language model that holds the firm’s judgment, so the system can’t improvise criteria or make commitments it can’t honour. The language model gives the words; Atrapos enforces the judgment.