business-brain-builder: the connected Brain

Setup Cowork-ready

What it does

Turns a new client into a connected, self-updating Brain. It wraps client-brain (the research and intake core) and adds the layer research cannot reach, by connecting read-only to the client’s actual accounts:

• Guided access and asset inventory, recording which surfaces you can reach and, critically, which account each one is by its platform-reported identifier.
• Expanded intake including a claims sign-off, so the owner formally approves what may be said.
• 24 months of real performance history, ingested read-only.
• Their own creative history, torn down through ad-forensic-teardown, so you know what already worked for this brand rather than what works in general.
• Lean LTV cohorts from order data.

Then it runs a gap interview and gets the owner to sign off on the roughly 20 facts everything else rests on, and sets up a weekly brain-sync plus a monthly brain-health report so it stays current instead of decaying.

Read-only against every connected surface. Brain writes are local files.

When to reach for it

• Onboarding a client whose accounts you can access. This is the full version.
• You want the Brain to stay accurate rather than becoming a snapshot of onboarding week.
• You need their performance history informing strategy, not category averages.
• Before a new-client launch, which calls this as its first step.

Where it runs

Cowork ✓ Desktop ✓

Both. Read access to ad accounts, analytics and store or CRM is what unlocks the extra layer.

What it needs from you

• Everything client-brain needs, plus:
• Read access to the ad accounts, analytics property, and store or CRM.
• The owner’s time for the gap interview and the sign-off. This is the step people skip and it is the step that makes the Brain trustworthy.

Where a surface is not reachable, you get a five-part setup message: what is missing, the exact identifier, where to get it, what this run degrades to, and the next best move. Not a shrug.

What you get back

• The full 29-module Brain, with every fact provenance-tagged by where it came from.
• 24 months of performance history in the Brain rather than in a spreadsheet nobody opens.
• Their creative history, analyzed.
• LTV cohorts.
• An owner-signed list of the load-bearing facts.
• A weekly sync and a monthly health report, so drift is visible.

A worked example

You: “Onboard examplebrand with account access.”

Connected: Meta ads, GA4, Shopify. Klaviyo not reachable, so it returned the setup message naming the exact env var and what stays NO-DATA without it. Ingested: 24 months of spend and revenue, 340 past creatives torn down. Finding surfaced at onboarding: their best-performing historical angle is one nobody on the current team knew about, last run 14 months ago. Owner signed off on 19 facts; 3 stayed gaps.

Tips & gotchas

• The account-identifier capture is the quiet safety feature. Recording exactly which ad account, pixel and property belong to this brand is what prevents a later audit producing a perfectly clean report about somebody else’s account.
• The sign-off is the point. Facts nobody confirmed become claims in client-facing copy eventually.
• Read-only means read-only. It never changes a setting in any connected system.
• Let the weekly sync run. A Brain that is not maintained is a Brain that quietly becomes wrong.

Related skills

Wraps client-brain. Uses ad-forensic-teardown, apollo and persona-architect. First step of new-client-launch.