Automation · Operations and reporting

Find what AI assistants get wrong about your brand and where it comes from, with Profound

An AI agent checks in Profound which false claims assistants make about your brand, which sources they cite, and sums it up on WhatsApp.

Show me this with my own operation See every automation

How do I know whether AI assistants are saying false things about my business?

If your company uses Profound, you write to your Met on WhatsApp something like "what are AI assistants getting wrong about us this month?". The Met finds your organization and the category you track, pulls the FactCheck score for the period and the list of inaccurate claims Profound found, reads the full answers where they show up and checks which domains the engines cite in that category. It replies with each false claim, the answer it came from and the sources being used, and leaves the summary as a document in Profound for your team. Fixing the source is your job: the Met does not change what the assistants say.

What starts it

When someone asks for it

It starts when someone on the team or at the agency asks the Met for the review on WhatsApp or in the internal chat. There is no trigger to configure: the Met runs the procedure in that conversation, with the category and period you give it.

A manual task: someone on the team runs it with a button, or asks the Met for it in chat.

Who runs it

AI search reputation analytics Met with Profound

This is not a named, seeded Met. It is an internal Met with Profound connected: it is done when it gives you the period's inaccurate claims with their answer and sources and leaves the summary written in Profound, or when it tells you which data it could not pull and why.

Run by the data analyst Met

Channels: WhatsApp · Website chat

What you need connected

Profound Research and analyze

6 steps · 7 tools

The procedure, step by step

Each step shows the tool that runs. They belong to the real connector: if one of them stopped existing, this page would not build.

1
Confirm the account and the organization

The Met confirms which Profound account it is connected with and which organizations and regions it can see. If the account has several organizations, it uses the one set in the skill or asks you which one.
- whoami Profound
2
Find the tracked category

The Met lists the organization's categories and picks your brand's, for example "Dental clinics in Bogotá". FactCheck and citation reports are requested by category, so without this step there is nothing to query.
- list_categories Profound
3
Pull the FactCheck score for the period

The Met requests the category's FactCheck scores for the date range you asked for, for example the last month. That tells you whether the accuracy of what is said about your brand went up or down compared with the previous period.
- get_factcheck_report Profound
4
List the inaccurate claims

The Met pulls the claims Profound flagged as inaccurate in that period: old opening hours, a price that no longer applies, a service you do not offer. It groups them by topic so you do not get the same claim repeated ten times.
- get_factcheck_claims Profound
5
Read the full answer and the cited sources

For the most repeated claims, the Met reads the AI answers Profound recorded and the category's citations report, and tells you which domains and pages the engines use. That is usually where the problem starts: an outdated directory or an old page on your own site.
- get_prompt_answers Profound
- get_citations_report Profound
6
Leave the summary written in Profound

The Met creates a Profound Docs document with each claim, where it showed up and which source to check. Creating a document with a name that already exists returns the existing one instead of a copy, so running the review again does not fill Profound with duplicates.
- create_doc Profound

Before you start

What you need ready

The Met solves none of these for you. If one is missing, the automation stops there.

Profound on the Enterprise plan, with your brand's category set up

Your company or your agency buys Profound and sets up the category and prompts it wants to track. Profound's documentation places the MCP server on the Enterprise plan.
Profound connected in Met

In Met, open Connections → Profound → Connect and sign in with an account that can see your brand's organization. Document creation arrives switched on; if you do not want the Met to write in Profound, the admin turns it off.
A channel to ask the Met for the review

The team's WhatsApp line or Met's internal chat, with the Met assigned. The answer arrives there; Profound stays as the source of data.

The limits

What this automation does NOT solve

Better said here than discovered halfway through the rollout.

It does not fix what AI assistants say. It tells you what is wrong and where it comes from; updating your site or asking a directory for a correction is your team's job.
It does not check prompts Profound does not track. If the question you care about is not set up in the category, it will not show up in FactCheck until your team adds it.
It does not work without Profound's Enterprise plan, according to its documentation. For a small business without that plan, this procedure does not apply.
It does not ask for confirmation in the chat before creating the document in Profound: if the action is on, it creates it. The admin decides which writes stay on.

FAQ

Questions about this automation

No. Profound measures what AI engines answer, and the Met only reads those measurements. What changes the answer is fixing the source behind it, and your team does that.

Whatever Profound flags in its FactCheck report for your category. The Met does not decide on its own what is false: it shows you what Profound found and the answer it came from.

Yes, if Profound tracks it. The Met lists the available engines and filters the reports by the one you ask for; if an engine is not on Profound's list, there is no data for it.

No. This procedure starts when you ask the Met for it, so the review happens each time someone on the team requests it with the category and the period.

Keep going