CoreCMO

Activate

Your context is written down. Now make every assistant read it.

Most teams get AI value one person at a time. Someone builds a good prompt, keeps it in a personal doc, and the value stops at the edge of their laptop. The reason team-level AI stalls is not model quality. It is that everyone is working from a different understanding of the company, so every output has to be corrected by hand before it can be used.

This page is the fix. It shows you exactly how to point the assistant your team already uses at one shared context, so the correcting stops.

You have two links. They do opposite jobs.

Once you claim your address in the Operator Brief, CoreCMO gives you two URLs. Sending the wrong one to the wrong audience is the only real mistake available here, so it is worth thirty seconds to get straight.

 Public pagePrivate context
The URL corecmo.ai/c/yourcompany
plus /brief.json and /llms.txt		 corecmo.ai/c/yourcompany/context.txt?k=<token>
What is in it Positioning, category, ICP, competitors, voice. Private fields such as KPIs, pricing internals, win rates and pipeline are stripped automatically. The complete operating brief, including the internal numbers the public page leaves out.
Who it is for Search engines, AI crawlers, and anyone researching you. This is the page that gets cited when a buyer asks an assistant about your category. Your own team and your own assistants. Nobody else.
How it is protected Nothing to protect. It is built to be found. Unguessable token, served noindex, nofollow and private, no-store, so it never enters a search index or an AI training corpus. A wrong or missing token returns a 404 that looks identical to "no such company."

THE ONE RULE

The private link is a password in URL form. Put it in your team's assistant, your password manager, or your internal wiki. Do not put it in a public repo, a shared slide, or anywhere a link preview bot will follow it. If it does get out, open your Operator Brief and click Rotate private link: the old URL dies immediately and you paste the new one wherever you had the old one.

Both links live in the My Companies menu in the top navigation once you are signed in, and in the publishing panel at the bottom of your Operator Brief.

Wire it into the assistant your team already uses

The pattern is the same everywhere: create a shared workspace, put the context in it once, and have everyone work inside that workspace instead of starting from a blank chat. Below is the specific path for each of the four tools we see most.

Claude

  1. Create a Project for your company. One project the whole team uses, not one each.
  2. Open Project knowledge and add your context. If your plan can fetch URLs, add the private context link directly so it stays current. If not, open the link, copy the text, and paste it in as a document.
  3. In the project's custom instructions, add: Answer as this company. Use the operating context in project knowledge for positioning, ICP, competitors and voice. Respect the confidence tag on every value.
  4. Invite the team to the project.

ChatGPT

  1. Create a Project (or a Custom GPT if you want to share it more widely inside your workspace).
  2. Upload the context file to the project's files, or paste it into the Custom GPT's Knowledge.
  3. Put the same instruction in the project instructions or the GPT's Instructions field.
  4. Share the GPT with your workspace so sales and customer success get the same answers marketing does.

Gemini

  1. Create a Gem for your company.
  2. Paste the context into the Gem's instructions, or attach it from Drive if you keep a copy there.
  3. Share the Gem with your team.

Microsoft Copilot

  1. Save the context file to the SharePoint or OneDrive location your team already has indexed.
  2. Reference that file in a Copilot agent, or point people at it directly in prompts.
  3. Because Copilot reads your tenant, keeping the file in one governed location matters more here than anywhere else. One file, one owner.

PASTED COPIES GO STALE. PLAN FOR IT.

If your tool can fetch the URL, it reads the current version every time and you never think about this again. If you pasted the text instead, that copy is a snapshot from the day you pasted it. Make replacing it the last step of your publish routine: whenever you promote a new version of the Brief, update the copy in each workspace. It takes a minute and it is the difference between a context layer and a document nobody trusts.

What to type once it is in there

You do not need special syntax. You need to stop re-explaining the company. Some examples of what changes:

Draft the competitive battlecard for our top three competitors, for an enterprise deal.
Write the onboarding email sequence for a new customer in our primary segment, in our voice.
Build next quarter's ABM plan against our ICP. Flag anything you do not have data for rather than guessing.
Turn our positioning into a one-page narrative a new sales hire could deliver in week one.

The assistant already knows who you sell to, what you are against, how you win, and how you sound. That is the whole difference. The same request that used to produce plausible category-average output now produces something recognisably yours.

The part that stops it lying to you

Every value in your context carries a confidence tag, and the file tells the model how to treat each one:

The instruction is explicit that confidence must not be upgraded in transit. This is the difference between grounding an assistant and simply giving it more material to be confidently wrong with. An estimate that travels into a board deck as a fact is worse than no number at all.

Keeping it honest as the team grows

A shared context only works if it stays one context. Three things hold that together:

Open your Operator Brief → Don't have one yet? Analyze your company
Strategic · Creative · Data Driven · Revenue Accelerator