Done-for-you implementation

Company Brain Sprint

The first usable version of your company-wide AI context layer — built with your team in 30 days.

Shared context, source-system boundaries, agent access points, optional runtime setup, reusable skills, QA gates, owners, and the rollout rhythm your team can keep using.

From £12,000 · Founding-client pricing for early teams in exchange for a case study · Two sprint slots per quarter.

For companies with AI activity, but no shared operating system.

Your company brain is the first layer of your AI operating system: approved business context your teams can reach from the tools they already use. The sprint builds that foundation in one focused area first, then proves it through real team use cases so it becomes infrastructure, not another AI experiment.

You have AI tools, but gains still depend on individual prompting skill

Important operating context lives in meetings, Slack threads, tickets, and old docs

Teams keep re-explaining the same background to Claude, ChatGPT, Copilot, or internal chat tools

You want your teams using the same approved business context, not each person building their own prompt stash

Leadership wants measurable output gains without adding equivalent headcount

You want a practical context foundation, not another strategy deck or point-solution automation

What happens in the sprint

Week 1

Diagnose the operating gap

  • Interview operators, team leads, and heavy AI users
  • Map the knowledge your people keep re-explaining: decisions, policies, workflows, customers, products, and edge cases
  • Inventory source systems, permissions, current AI usage, and never-ingest boundaries
  • Choose the first team use cases that will prove the brain is usable

Week 2

Design the operating layer

  • Define the context architecture, trust metadata, and decision rules
  • Design the first brain structure, reusable skills, and write-back flow
  • Set owners, QA gates, and approval rules
  • Design the access and runtime pattern: which agents run in existing tools, which require Slack/Discord/CLI/webhook/cron surfaces, which tools they can call, and what must stay human-approved

Weeks 3–4

Install the first version

  • Build the first context packs, access points, skills, instructions, and — where the use case requires it — the first lightweight runtime surface
  • If appropriate, set up a company agent runtime using Hermes or another agreed pattern, scoped to the pilot workflow rather than a broad platform rollout
  • Test the brain live through real team use cases
  • Capture failures, tighten the system, and remove friction
  • Leave the context foundation, ownership rhythm, and rollout pattern your other teams can copy

Runtime setup, when useful

The brain can also run through a persistent company agent.

Some teams only need better access from Claude, ChatGPT, Cursor, or Copilot. Others need an agent that sits in Slack or Discord, runs on a schedule, listens for webhooks, works from the CLI, or executes recurring operating loops. In those cases, the sprint can set up the first runtime pattern around the Company Brain.

Surfaces

Slack, Discord, CLI, webhooks, cron, or internal tools — chosen around where work already happens.

Context loading

The runtime loads approved brain documents and shared skills before it acts.

Tools and actions

MCP servers, APIs, and connectors let the agent retrieve live facts or take approved actions.

Human gates

Sensitive actions, customer impact, financial changes, and brain updates go through explicit review.

Optional Hermes path

Hermes can be used as an open-source runtime if it fits your stack; otherwise we use a lighter agreed pattern.

The sprint does not turn 10xyourteams into a generic automation agency. Runtime setup is scoped to proving and operating the Company Brain.

Proof it works, concretely

The brain proves itself in live team use cases.

The goal is the shared context foundation. Use cases are how we test it: someone asks, the agent answers with approved company context, and what’s accepted flows back in with review. Three examples:

“This escalation mentions a chargeback — what do I do?”

What comes back: The matching policy, this customer's history, and a drafted reply that respects the refund threshold.

What writes back: The resolution is proposed to the decision log, so the next escalation starts smarter.

“Why is contribution margin down this week?”

What comes back: The variance against plan, the three movements driving it, and the invoices or campaigns behind each.

What writes back: The explanation lands in the weekly digest — leadership stops re-asking in three meetings.

“What needs action in merchandising today?”

What comes back: Stockouts, sell-through outliers, and suggested markdowns or transfers — with the reasoning shown.

What writes back: Accepted actions are recorded with owners; rejected ones teach the system your judgement.

What you get

Company context map across source systems, teams, decisions, workflows, and permissions

The first approved Company Brain structure: what lives in the brain, what stays in source systems, and what is never ingested

Approved context architecture and source-of-truth model

Reusable skills, instructions, and example use cases that prove the brain works

Agent access points and optional runtime setup: MCP, APIs, integrations, Slack/Discord/CLI/webhook/cron surfaces, and a permission map for where AI can safely work

Runtime operating model: which agents run where, which brain context and skills they load, which tools they can call, which actions require approval, and how proposed brain updates are reviewed

QA, review, and approval gates for AI-assisted output

Named owners, freshness rules, rollout rhythm, and adoption cadence

Team enablement session

30/60/90-day scale-up plan

The internal case

Scoped around payback, not novelty.

No claims that AI will magically grow revenue. The business case starts with cost avoided: repeated coordination, manual reporting, rework, duplicated context, slow handoffs. We prove the brain through practical use cases, then the margin comes from each team that starts from the same context foundation instead of rebuilding it alone.

We choose visible use cases where saved hours are measurable — not demo-first use cases

We measure before and after: manual hours, delay, rework, escalation volume

We count cost avoided first; revenue upside is a bonus, never the pitch

You leave with the payback maths your CFO will accept, filled in with your numbers

After the first Sprint

The first brain is the pattern. The value compounds when other teams copy it.

The 30-day Sprint is not meant to solve every workflow in your company. It installs the first usable Company Brain, proves it through real work, and leaves the operating pattern your other teams can reuse.

Roll it out yourselves

Your team owns the brain, templates, access setup, review rhythm, and rollout plan. Most companies can expand from the first team into the next one using the Sprint pattern.

Add another team or operating loop

Use the same structure to bring in support, product, operations, finance, trading, customer success, or engineering without rebuilding the system from scratch.

Keep Marc involved as Momentum

Monthly review, new use-case selection, context-quality checks, skill updates, adoption support, and the next layer of runtime/workflow design.

The goal is not dependency. The goal is a Company Brain your team can keep improving without waiting for another consultant.

Boundary

Not a workshop. Not a prompt pack. Not a custom automation shop.

The sprint installs the operating layer AI depends on: shared context, source-system access, reusable skills, ownership, and review loops. Specific workflows and tools can change. The system stays useful.

Not a fit if you want:

  • ×Generic AI awareness training
  • ×A prompt library without workflow change
  • ×A vendor to automate every process indefinitely
  • ×To skip naming an internal owner for context quality and upkeep
  • ×Tool adoption metrics without changing how work runs

Questions leaders ask

How is this different from Glean, Notion AI, or enterprise search?

Search tools index whatever exists — including the stale and the contradictory. A company brain is curated: every document is owned, dated, and reviewed, agents reach it through controlled access points, and approved learnings flow back in. Search finds documents; a brain gives agents operating knowledge they can act on. The two can coexist — search is an input, not the system.

What about our data and security?

The brain lives in your own private repository. People and agents reach it through one SSO-authenticated access layer with per-user permissions and a full audit log — and we agree a never-ingest list (personal data, legal, security material) before anything is captured. It is a narrower, more controlled surface than the document sprawl most companies run on today. Your security team gets the full design.

How much of my team's time does the sprint take?

Leadership: a kickoff and a short weekly check-in. The pilot team: interviews in week one, then doing their normal work the new way in weeks three and four. The heavy lifting in between — design, writing, setup — is mine.

We're not a very technical company. Will this work?

Yes. Most of the work is making operating knowledge explicit, which is an operator's job, not an engineer's. The technical layer is small and I handle the setup. Your team uses the result through tools they already know: Claude, ChatGPT, Slack.

Can the sprint set up Hermes?

Yes, if a persistent runtime is the right answer for the pilot. Hermes is one open-source option for running company agents across chat, CLI, webhooks, cron, and tools. But the sprint starts with the Company Brain. Sometimes the right runtime is Hermes; sometimes it is an existing Slack bot, a lightweight service, or simply better access from the tools your team already uses.

Is Hermes required?

No. The durable asset is the Company Brain: context, skills, governance, and write-back. A runtime operationalizes that brain. Hermes is an optional implementation choice, not the product.

What happens after the 30 days?

You own everything: the brain, the templates, the access setup, and the rollout pattern other teams copy. Most teams continue on their own. If you want ongoing help, there is an advisory option to keep the system improving as teams and tools change.

How soon can we start?

I take on two sprints per quarter, alongside leading AI implementation inside a 1,000-person commerce group. Book a call to check the next opening.

Get a company brain your agents can use.

If it works once, the rest of your company has a pattern to copy. From £12,000.