BBT AI.
Your brand is now being generated in places you will never see, by teams, partners, platforms and models. BBT AI is how it stays yours. We scan where it has already drifted, then write it in a form machines can follow.
What usually goes wrong.
None of these are technology problems. They are all problems of a brand that was never written down properly.
- 01Every team can now generate brand content, and nobody agrees what is on brand.
- 02The guidelines are a PDF. The work is being made by a model that cannot read it.
- 03The output arrives fast, looks fine, and sounds like everybody else.
- 04Nobody can say who approved what the machine made.
Five modules. Start with the one the problem needs.
BrandScan tells you how far the brand has already drifted. BrandCode is what holds it together afterwards. Most clients start with the scan, because it is the cheapest way to find out whether the rest is needed.
BrandScan
Where the brand has already driftedBefore anything gets encoded, we look at what is actually out there, and at what the machines are already saying about you. It runs on its own, as a first engagement, and often should.
- Touchpoint scanEvery place the brand shows up, measured against what it is supposed to be.
- Machine scanWhat a customer gets when they ask an AI about your category and your company.
- Drift reportThe gaps costing you recognition, trust or a sale, ranked so somebody can act.
- The case for the codeWritten for the person who signs the budget, not the one who noticed.
BrandCode File
Written so a model can follow itStrategy, identity, voice and rules turned into structured intelligence a model can act on, and a person can still read and argue with.
- Positioning and narrativeThe reason to buy and the reason to care, in a form that can be applied.
- Voice and vocabularyHow it sounds, and the words the brand refuses to use.
- Identity rulesColour, type, imagery and layout logic a generator can obey.
- Worked examplesRight and wrong, side by side, because rules alone are not enough.
BrandCode QA
Drift caught at the draft, not the launchWork run against the code first by machine and then by a person, so the conversation happens while it is still cheap to change.
- Automatic checkEvery piece of work measured against the code before anybody sees it.
- Flags that matterRanked by consequence, so nobody is drowning in trivial notes.
- A human decisionAnything borderline goes to a person, with the reason attached.
- A recordWhat was approved, by whom, and why, kept for the next time it comes up.
BrandCode OS
One versioned source, everywhereWhere the code lives, versioned, so your teams, your agencies and your platforms are working from the same brand on the same day. It is also where the code gets wired into the tools people already open.
- One source of truthVersioned, so there is never a question of which file is current.
- Access for everyoneTeams, agencies, platforms and models, each with the right level.
- Wired into the toolsThe on-brand answer becomes the quickest one to reach.
- Updates that landA change reaches every partner at once, not eventually.
Governance
Including what the machine may decideWhat a model may decide on its own, what needs a human, who signs it, and what happens the day it gets something wrong in public.
- The machine's laneRoutine adaptation, resizing, versioning. Decided without asking.
- Where a person signsClaims, campaigns and anything public the brand has not said before.
- The escalation pathNamed people, an agreed clock, and what gets pulled first.
- Review as it movesThe rules revisited as the business changes, not left to rot.
Once the code exists, the work gets made against it.
Image, motion, film, content and adaptation, produced at AI speed and signed off by people. Not a separate service. It is what Build looks like once the brand is encoded.
The code is what makes this safe to do at volume. Without it, speed is only a faster way to drift. See Build →
Ten weeks to set up. Then it runs.
Run by the same senior people who resolve the brand itself. Three of these four have no end date, which is the point.
Every touchpoint and every model answer, gathered and measured against what the brand is supposed to be.
Strategy, identity, voice and rules written as structured intelligence, with worked examples.
The checks, the approval path, and what happens the day the machine gets something wrong.
The code into the tools your teams and partners already use, with the people trained on it.
Scoped per SOW. Set up once, then kept running.
BrandScan is quoted on its own and can be the whole first engagement. The set-up is a defined scope. The scan cycle, QA and the OS continue as an ongoing engagement, scoped to the estate they cover.
BrandScan and BrandCode open shortly.
The modules are being built. This page opens when they are ready.
Start a conversation →