Skip to content

Let your agent set it up

Connect your AI to the Omnipresence MCP and have it do the setup: the brand, the competitors and the buyer prompts, from one starter prompt.

Updated October 20, 2026

This is the fastest route in, and the one the setup checklist leads with. You connect your own assistant to Omnipresence over MCP, give it one prompt, and it fills in the brand, the competitors and the first set of buyer questions by asking you about the business rather than making you fill in a form.

Why this is the main path

The setup that matters is judgment, not typing: which competitors you actually meet, how buyers phrase the question, which spellings of your brand show up in the wild. An assistant that can see your site and ask you three questions gets there faster than a form can, and it runs on your own AI subscription, so the exploration costs you nothing here.

Do it

1. Connect the MCP

Add Omnipresence as an MCP server in your client. The endpoint is the same for everyone:

https://api.getomnipresence.com/mcp

In Claude Code that is one line:

claude mcp add --transport http omnipresence https://api.getomnipresence.com/mcp
bash

Claude.ai, ChatGPT, Cursor and Claude Code all have a guided version of this in the project's MCP tab. Sign in when the client opens the authorisation window. Connecting your agent has the detail, including what to do when a client does not finish the handshake.

2. Send the starter prompt

Paste this, exactly:

Run the Omnipresence MCP onboarding

The checklist on a new project has a copy button for it, because the wording matters: it is what tells the agent to fetch the onboarding process from the methodology rather than improvising a setup of its own.

3. Answer its questions

It will ask for the domain, then about what you sell, who buys it and who you lose deals to. Answer in plain sentences. "We sell scheduling software to independent dental practices in the UK, and we usually lose to the two big practice-management suites" is more useful than a tidy list, because the phrasing is what the buyer prompts get built from.

4. Check what it wrote

The agent writes real configuration, so read it back:

  • Settings: brand name, name variants, domain, vertical, market.
  • Name matching: your brand and each competitor, with their alternate spellings.
  • Prompts: the tracked prompts it created, with a cadence.

Everything is editable. Change anything that reads wrong, and delete any prompt that does not sound like a question a buyer would type.

5. Let the schedule run

Nothing else is needed. The prompts run on their cadence, and the checklist takes itself off the project once the first runs land.

Tip

Ask it to explain its choices: "why those five competitors, and why those prompts?" The answer tells you whether it understood the market, and it is much faster to correct the reasoning now than to re-read a month of data built on the wrong competitor set.

What it cannot do for you

Add your model key. That is a credential, so it stays a deliberate human action in the app. If you have not added one yet, do that first, or the agent will set up a project that cannot run anything.

On a Starter plan the MCP connection is read-only, so the agent can look but not write. The onboarding path needs the write tools, which start at Pro.

Questions about let your agent set it up

Does this spend my model budget?
The setup conversation runs on your own AI subscription, not on Omnipresence. Only the tracked prompts themselves run on your model key.
Can I change what the agent sets up?
All of it. Everything it writes is ordinary project configuration, editable in Settings and in Name matching afterwards.
What if I would rather do it myself?
Set it up by hand instead. The manual path reaches the same place and the checklist offers both.

Related

Was this page helpful?

Not tracking yet?

Seven days free, on your own model key. Point it at one brand and read what comes back.

Start your free trial