01
It starts from nothing
Every question begins by re-explaining the business.
AI adoption for UK businesses
Take on more work without taking on more people.
Your business already produces the evidence of how it works: emails, documents, records, transcripts. What it has never produced is the reasoning behind any of it. We read the evidence, interview the people who hold the reasoning, and map the two together into one brain your systems and your AI can act on.
Baobab NetworkCognifraxisForas GroupThree Peaks HealthWorker BeeDentsuDAZN
Every line is a handoff · hover a point
The problem
AI does not know how your company works.
Everyone has the same models. None of them have seen your rates, your standards, or the reason behind a single decision you have ever made. So you get answers that sound right and are wrong for you.
02
It answers generically
Well written, wrong for you. Someone fixes it by hand.
03
Nothing is kept
What one person works out never reaches anyone else. Every month starts again.
The answer
The evidence and the reasoning, mapped together.
One half is objective and already written: every email, document, record and transcript your business has produced. The other half is subjective and never written: why you priced it that way, why you said no, what a good job actually looks like. Neither is much use alone. Mapped to each other, they are a brain.
Your systems
XeroAccounts
HubSpotCRM
GmailEmail
Google DriveDocuments
TrelloProjectsThe evidence · whatever you already run on
Your company brain
FilteredCitedYours
Your people
Sales leadPricing & discounts
OperationsHow jobs get scheduled
FinanceMargin floors & terms
DeliveryWhat good looks like
Client servicesWhat clients always ask
The reasoning · everyone who holds a rule
What the brain keeps
- Process, policy, rates, standards, templates
- The rules in your experts’ heads that were never typed out
- What jobs actually cost and how long they took
- A citation on every one of them, back to where it came from
What it reads but does not keep
- Raw email threads and meeting transcripts. Read as evidence, cited, not stored as knowledge
- Personal details, stripped before anything is modelled
- Anything that stops being true within three months
What it can do
It assists your team. Or it does the work itself.
Two jobs, from the one structure.
01It assists your team
They ask it something, or they tell it to make something. Either way it answers from your written rules, and shows which ones it used.
Ask it“Can we take the £40k job in March?”
SalesTwo like it, at £38k and £44k
DeliveryThree engineers, five weeks
OperationsTwo of them are away in March
FinanceMargin floor is 38%
XeroThe last one ran 12% over
TrelloSimilar jobs took six, not five
It answersYes, starting 17 March, at £44k not £40k.
Ask itPricing
“What should we quote for this?”
What the last four like it sold for, and what they actually cost to deliver.
Ask itCapacity
“What will we have to turn away next quarter?”
Work already committed, against the people who can actually do it.
Ask itClient history
“Why did we agree to that with this client?”
The decision, the thread behind it, and the document that recorded it.
Tell itProposals
“Draft the proposal for this enquiry.”
Back in your format, at your rates, with the margin already checked against your floor.
Tell itReporting
“Build the pack for Thursday’s board meeting.”
Pulled from the systems, in the layout you use, with last month’s commentary brought forward.
Tell itCompliance
“Write this up so it passes a file check.”
Against your own standard, with the reasoning recorded the way your reviewer expects to see it.
02It automates the work
Nobody asks. Something happens, and the work is done.
A person approvesA new enquiry lands
The proposal is drafted from the three closest jobs you have won, at your rates, margin already checked.
Sales lead reads it and sends itRuns on its ownMonday, 7am
The pipeline report and the cash position are already built. Nobody rebuilt a spreadsheet.
Logged, and reviewable afterA person approvesA project closes
The billable lines, the retro and the case study are written up from the record.
Delivery lead approvesA person approvesA job starts slipping
The account owner is told that week, with what changed and what it cost. Before the client says anything.
Account owner decidesA person approvesA client asks a repeat question
The reply is written, using that account’s history and your policy. Ready to send.
Support checks and sendsRuns on its ownSomebody new starts
They are briefed on how their job actually works, and can ask it the things they would rather not ask twice.
Logged, and reviewable afterYou decide which of these run on their own and which wait for a person.
Case study
A regulated financial services firm.
£1m revenue. Turning work away and about to hire for it.
The costed breakdown
Their own map, with the company name removed. Click a division to read every process in it, the hours it takes today, and whether it gets automated, drafted for approval, or left with a person.
94
Processes
13
Functions
~120h
Hours back a week
~£201k
A year back
HOURS × RETURN FACTOR × £35/HR × 48 WEEKS | AUTOMATED 0.8 · DRAFTED 0.5 · TRAINED 0.2
All thirteen divisions and all ninety-four processes, as they were mapped. Division totals are theirs; the split of hours inside each division is modelled to them. Illustrative estimates, not forecasts.
We automate it
It runs without anybody starting it. Only where the rules can be written down and the result checked afterwards.
AI drafts, your team approves
We build the part that produces the work. Your person reads it, changes what is wrong and sends it.
We train your team
The job needs judgement or a relationship, so it stays with the person. We teach them to do it with the brain behind them.
The map is yours at the end of month one, whatever you decide next. Builds are quoted against it, cheapest payback first.
Customer references.
Baobab NetworkAfrican venture investor
Forty thousand applications a year, screened by hand. We built their own investment judgement into a system that does the first pass, then checked it against them.
Agreement with human decisions86%
Investable companies missedzero
Cost per application−90%+
Cognifraxis Medical CenterPrivate practice, 22 clinicians
Three sites running on five separate tools, twenty spreadsheets and paper diaries. One platform replaced the lot, built to feel like Excel because the clinicians had refused everything else.
Annual revenue€1.08m → €1.39m
Capacity utilisation65% → 83%
Admin hours released450+ a month
Three Peaks HealthLongevity clinic
Clinicians spent hours assembling patient data before they could apply any judgement to it. Bloodwork, lifestyle data and lab results now arrive in one view, before the appointment.
Clinician pre-analysis time−50–70%
Full patient dashboardunder 2 min
Patient dataOne view
“This has fundamentally improved how quickly we can move from raw data to insight. We’ve cut pre-analysis time dramatically and can deliver more holistic recommendations with greater confidence.”
“By embedding intelligent automation into our workflows, Follow the Rabbit gave our teams the space to push creative boundaries without being slowed down by the operational grind.”
“Follow the Rabbit combines clear focus with strategic thinking, aligning with our needs while innovatively pushing boundaries to drive impactful results.”
“Rabbit helped change the way we thought about driving retention through their consultancy and psychological framework. Working with them we saw a 90% user return vs. rate alone.”
Worker Bee (Banijay UK) 30%+ fewer pickup shoots · Dentsu 40% faster content delivery · Unidays 90% user return rate · DAZN AI roadmap used to choose between pilots
Two ways to start.
Half an hour on a call. You will leave knowing which division to write down first, and roughly how many hours a week are sitting in it.