Setting up a consulting company’s work on AI

A new consulting company had to research its market, build its target accounts, write its outreach, price its services and launch its website, all at once. We set the work up on Claude and ChatGPT with one shared set of rules, a workspace for each job and a written handoff for every task.

Consulting companyGetting started with AI

What changed in the first week

These counts come from the company’s own working files.

224Target accounts researched against the ideal customer profile, 29 of them in full dossiers
18Industry and platform pages drafted, with a source behind every claim

Where they started

The company was new, and research, sales, pricing, content and the website all needed doing at the same time. Without shared rules, every AI session started from nothing, and the same instructions had to be given again for each task.

What we built

We set up the assistants the company already paid for and gave every job the same rules and a workspace of its own.

  • A Claude Ultra plan for the main work and ChatGPT for image work.
  • One Claude Project holding the brief, the writing rules, an evidence standard and the sales method, which every session reads first.
  • A workspace for each job, such as research, accounts, outreach, pricing, content and brand, each with its own decision log and evidence log.
  • A written handoff for every task, so a new session picks up exactly where the last one stopped.
  • Claude Code for every build and Claude Design for every design, with each design approved before any code.

How it runs today

Every task starts from a written handoff, every claim in the work carries its source, and the owner approves each decision before it ships.

Inside the workflow

Select any part of the map to see what it reads, what it writes and who approves the result.

Shared setup

One set of rules, a workspace for each job and a written handoff for every task, across Claude and ChatGPT.

Runs onEvery task in every workspace
Writes toEach workspace’s files and logs
Approved byThe owner, before anything ships
Moving lines show which way information flows

How the workflow runs

1

A handoff is written

Each task starts from a written handoff with its goal, the files to read and the checks.

2

The rules come first

The session reads the shared rules before it writes anything.

3

The work is logged

Each decision goes into the workspace’s decision log with its source.

4

The owner approves

The owner reviews the result and approves it before it ships.

More case studies

See all case studies

Bring us the AI your team uses on its own

Start with a working session where we watch it with your team, and the workflow map that follows ends in a fixed quote.

Book a working session