Client delivery brain for agencies

Admin Chief for agencies

Your client history is scattered across proposals, call recordings, Notion, and individual inboxes. We organize it into a structured brain, then run Scout, your Client Context Retriever, so your team has the right context before every call, report, and strategy session.

Client Delivery Brain (sample)
Sources to wiki to agent output
Sources ingested
Past proposalsGoogle Drive, 40 plus docs
Call transcriptsFireflies, 60 summaries
Client notesNotion, 3 workspaces
Campaign reportsPDF and Sheets
SOPsNotion, 12 docs
Structured wiki
ClientsStatus, contacts, contract terms
ProjectsGoals, deliverables, outcomes
CallsDecisions and action items extracted
ProposalsOffer, scope, price reasoning
SOPsReusable, tagged by function
Skill output
Scout: Client Context Retriever
Pre-call brief: Hartwell Studio (sample)Last proposal: brand refresh, March. Decision: Q3 launch. Open item: final asset sign-off pending. Campaign note (sample): email series delivered, open rate tracked at 18% above baseline.

Sample prototype layout. Not real client data. No live client system is implied.

Where agency context goes wrong.

The information exists. The problem is that it cannot be reached quickly, and nobody has the job of keeping it current.

Where it lives now

  • Proposals buried in Google Drive folders named by whoever sent them last.
  • Client history split across Notion, Slack, and a call transcript nobody reviewed after the meeting.
  • Campaign results in a report that was emailed to the client but never filed back into the client record.
  • SOPs written once and never updated when the process changed.

What it costs the team

  • Account handlers ask the same questions that were answered six months ago in a proposal.
  • New team members spend their first weeks excavating context instead of delivering work.
  • Pre-call prep takes 30 minutes of searching when the brief should take five minutes to read.
  • Context walks out when people leave, because it was never written down outside of their heads.

What the client delivery brain is built from.

The brain ingests your existing sources. Nothing new to create: we work with what you already have.

Past proposalsScope, pricing, and reasoning from every pitch
Call transcriptsDecisions and open items extracted from recordings
Client notesRelationship context, preferences, and concerns
Campaign reportsResults, commentary, and delivery history
SOPsProcess knowledge tagged by function and team

Five stages. One operating brain.

Built and maintained on our own infrastructure. Each stage has a clear role: nothing sits idle.

01

Ingest

Pull in proposals, transcripts, notes, reports, and SOPs from the agreed sources. Format and access are mapped at the Context Audit.

02

Structure

Organize everything into an AI-readable wiki with a consistent schema: clients, projects, calls, proposals, and SOPs each have a defined shape.

03

Retrieve

Surface the right context on demand: the right client record, the right proposal excerpt, the right call decision.

04

Serve

Run Scout, your Client Context Retriever: reads the brain and produces a pre-call brief for your team to review before acting on it.

05

Maintain

Scheduled freshness sweeps keep the brain current as new proposals, calls, and reports come in. Nothing is left to rot.

Flagship agent: Scout, your Client Context Retriever

Scout is your prep-ahead chief of staff. Before a call, report, or strategy session, Scout surfaces the relevant client history from the brain and produces a one-page brief for your team to review. Signature artifact: The 60-Second Brief.

What Scout reads

Scout, your Client Context Retriever, pulls from the structured client record: past proposal scope and price, call decisions and open items, campaign results, and any relevant SOP references. Scout does not search the raw documents; it reads the organized wiki where the signal has already been extracted.

What your team receives

The 60-Second Brief: one document, organized by section, ready for a human to review before the meeting. The team confirms it is accurate, adds anything recent, and uses it. Nothing goes to the client directly.

Output reviewed and approved by your team before any use. Approval-first is a firm operating rule.

Scout (Client Context Retriever)
Sample output only
SAMPLE ARTIFACT

Pre-call brief: Hartwell Studio

Prepared from brain. For team review before the 14:00 strategy call.

Last proposal

Brand refresh and content system, March. Agreed scope: visual identity, 3 page templates, quarterly content calendar. Price band (sample): mid-range retainer. Status: active.

Last call decisions

  • Q3 launch confirmed, design assets due end of June (sample)
  • Client requested fortnightly check-ins, not weekly
  • Open item: final asset sign-off from their head of brand still pending

Recent result

Email series delivered in May (sample). Open rate tracked above the agreed benchmark. No complaints raised. Client flagged copy tone as "on point" in the follow-up thread.

Watch

Budget review due at the end of this quarter per the contract note (sample). Worth confirming renewal intent on this call.

Admin Chief organizes and operates your company context. It does not provide legal, financial, or tax advice. No client-facing message is sent without your approval.

Three steps. Context Audit first, always.

The Context Audit maps your sources before any build quote is written. That is how fixed prices stay fixed and scope stays honest.

Step 2

Brain Build

£2,000 to £5,000

One client delivery brain built from the agreed sources, with the schema applied and Scout (Client Context Retriever) stood up behind an approval gate.

  • Scope capped to sources agreed at audit
  • AI-readable wiki structured and organized
  • Scout (Client Context Retriever) built and tested
  • Approval-first: output reviewed before use
  • Priced by source count and volume, not flat
Step 3

Maintain and Operate

£2,000 /mo

Scheduled freshness sweeps keep the brain current as your client base changes. Skill tuning and one new or upgraded skill each quarter.

  • Scheduled freshness sweeps
  • Skill tuning against real output
  • One new or upgraded skill per quarter
  • Monthly review against output value

Common questions.

What counts as a source?

Any document, recording, or structured export your team already has: Google Drive folders, Notion workspaces, Fireflies exports, campaign PDFs. The Context Audit maps what you have and grades the format before we quote the build. Messy sources cost more time and that cost is explicit, not hidden.

Does the brain replace our Notion or Drive?

No. The brain reads from your existing tools and organizes a structured wiki on top. Your team keeps working in Notion and Drive. The brain is the organized layer that makes those sources AI-usable and skill-ready.

What does "approval-first" mean in practice?

Every output from Scout (Client Context Retriever) is a draft for your team to review. No brief, no note, no draft is used until a human on your team has read it and decided to act on it. Nothing goes to a client without that step. This is a firm operating rule, not a disclaimer buried in the small print.

What is the Context Audit?

A paid scoping engagement (500 to 750 pounds) where we inventory your context sources, map their format and condition, propose a schema and organization rules, scope Scout (Client Context Retriever), and write a fixed build quote. It is the only route to a defensible price, because data-loading effort is the main variable that blows up flat quotes.

Paid scoping engagement

Book a Context Audit.

We map your sources, propose a schema, scope Scout (Client Context Retriever), and produce a fixed build quote. That is the whole audit.

Book a Context Audit