Sources AI relies on
- g2.comListed
- forbes.comNot listed
- reddit.com/r/HVACNot listed
- capterra.comListed
AI visibility for SEO and GEO agencies
Find out who AI recommends instead of your clients.
Mentiary asks ChatGPT, Gemini, Claude, Perplexity, and Google’s AI Overviews the questions your clients’ buyers ask. You get a white-label report with share of voice, the sources AI trusts, and the pages to get onto next.
Share of voice · Sample
Northline
75 AI answers
Housecall Pro62%
Jobber58%
Northline34%
ServiceTitan30%
Northline by engine
Google AI
40%
ChatGPT
30%
Gemini
50%
Claude
20%
Perplexity
70%
Every report covers
- Google AI Overviews
- ChatGPT
- Gemini
- Claude
- Perplexity
How it works
One loop, from cold prospect to retained client.
The same measurement runs at every stage, so the gap you pitch is the gap you report on, and the gap you close.
- 01Prospect
Audit a whole lead list
Upload a CSV of prospect websites. Each row comes back with the search where AI leaves them out, the competitor it names instead, and a first line for your cold email. Cheap enough to run on thousands of rows.
- 02Pitch
Send warm replies a full report
Every prompt is asked three times on five engines. Share of voice, prompt-by-prompt results, the sources AI cites, and what it gets wrong about the brand, in your branding.
- 03Retain
Show progress every month
The same prompts run again monthly or weekly. Clients see each engine move over time, and you see which work paid off.
The report
Built to be read by a client, defended by an agency.
It opens with one number a client understands: how much of their market’s search demand ends in an AI answer that names them. Then it shows the prompts behind it, the competitors named instead, and what to do about it.
Prompt by prompt
Sample data for a fictional brand · 3 runs per engine| Prompt | Google AI | ChatGPT | Gemini | Claude | Perplexity |
|---|---|---|---|---|---|
field service scheduling software 4,400 searches a month | Named | Sometimes | Named | Not named | Named |
best hvac scheduling software 2,900 searches a month | Not named | Not named | Sometimes | Not named | Named |
hvac dispatch software 1,900 searches a month | Sometimes | Named | Named | Not named | Named |
servicetitan alternatives for small shops 1,000 searches a month | Not named | Not named | Not named | Not named | Sometimes |
What AI says about the brand
“Northline starts at $99 a month per technician.”
Wrong. The pricing page says $79. Checked against the client’s own site.
Next step, drafted
Ask to be added to “HVAC scheduling software compared”.
ChatGPT cites it for 2,900 monthly searches. The note to the author is already written.
Made for agencies
Numbers you can put your name on.
Your brand on every page
Your logo, colors, and agency name on the PDF. Mentiary never appears in what your client sees.
Real answers, not guesses
ChatGPT and Gemini answers are collected as users see them, with their sources. Claude and Perplexity answer with live web search.
Measured more than once
AI answers change between runs. Each prompt is asked three times, and a brand counts as named when it appears in most of them.
Opportunities with drafts
Pages to get listed on, open Reddit threads, and pages to publish, each with a ready draft your team can edit and send.
Weighted by real demand
Prompts are tied to monthly Google searches, so a gap on a big search counts for more than a gap on a niche one.
Your prompt list, kept steady
Edit what gets tracked. Monitoring reuses the same list, so this month compares like for like with the last.
Pricing
Pay for reports and clients, not seats.
Your whole team gets in. Founding agencies lock in early pricing before public launch.
Get early accessProspect audits
per row
Run on cold lists. Rows where AI already names the prospect are flagged, so you only pitch real gaps.
Full reports
report credits
Five engines, three runs per prompt, opportunities with drafts. For warm replies and new clients.
Monitoring
per tracked client
Monthly by default, weekly when a client wants it. Capped per plan, so there are no surprise bills.
FAQ
Questions agencies ask.
Which AI engines do you track?
Google AI Overviews, ChatGPT, Gemini, Claude, and Perplexity. Every report covers all five unless you choose fewer.
How do you decide which prompts to track?
We read the client's site, write the questions their buyers ask an AI assistant, and keep the ones with real search demand, spread across topics. You can edit the list before the first run and at any time after.
Why ask each prompt three times?
AI answers vary from one run to the next. One run can make a brand look better or worse than it is. Three runs per engine, counted by majority, gives numbers you can defend to a client.
Is it white label?
Yes. Reports carry your agency name, logo, and colors. Your clients never see Mentiary.
Do you guarantee AI mentions?
No one can. We show where your client stands, which sources the answers rely on, and the concrete steps most likely to change that. The monthly report shows whether it worked.
Which markets do you support?
We start with the US. Other countries and languages work too, because prompts and search volumes follow the client's market.
Be the agency that already knows what AI says.
We are onboarding a small group of agencies first. Tell us a little about yours and we will reach out.