There's an uncomfortable test we put off for a while: open ChatGPT and Perplexity, ask them about your own business, and see what comes back. We're an AI agency. We assumed we'd look fine.
We didn't. When we ran our own ChatGPT-Visibility Kit against promptish.io, the technical audit was blunt: nothing structured for an AI to read. No machine-readable identity, no quotable content, no entity links. By our own kit's standard, we were close to invisible to AI — the exact gap we tell clients to close.
Good news for this post: that makes us the most honest possible test case. Here's what we found and exactly what we did.
The audit: what we found (the "before")
| Check (from our kit) | Before |
|---|---|
| Structured data (JSON-LD) | ✗ none (0 blocks) |
| Organization schema — machine-readable identity | ✗ missing |
| FAQ content + FAQPage schema (quotable) | ✗ none |
| Clear "what we do / who for" in plain words | ~ vague ("digital alchemy") |
| Canonical URL | ✗ missing |
An AI assistant names businesses it can identify, understand, and corroborate. We'd given it almost nothing to work with on any of the three.
The four fixes we shipped
- Organization schema — a machine-readable identity block: who we are, what we do, what we specialize in, in plain language an AI can lift verbatim.
- FAQPage schema + a real, visible FAQ — the four questions a prospect actually asks ("what do you do?", "who's it for?", "can you help me show up in AI search?", "how do I start?"), each answered in a self-contained way an assistant can quote.
- A plain "what we do" description — swapped poetry for a sentence a model can parse: an AI and software agency that builds custom AI systems, automation, and web products.
- A canonical URL — one definitive address so the entity isn't ambiguous.
Every one of these is copy-paste structured data and content — no rebuild, no redesign. It's exactly what's in the kit.
The result (the "after")
| Check | Before | After |
|---|---|---|
| JSON-LD blocks live | 0 | 2 |
| Organization schema | ✗ | ✓ live |
| FAQPage + visible FAQ | ✗ | ✓ live |
| Canonical | ✗ | ✓ |
The structured-data before/after above is immediate and verifiable — you can view-source promptish.io right now. The thing it's for — AI assistants describing and naming us better — takes weeks, because the platforms re-index on their own schedule. So we're not claiming "ChatGPT now recommends us." We fixed every input in our control; we'll measure and report the output honestly as it lands. That's the same promise the kit makes: no guarantees of placement, just the controllable parts done right.
The takeaway for your business
If an AI agency's own site had these gaps, yours very likely does too — and that's good news, because they're all fixable in an afternoon. Run the free 10-minute audit on your own business first. If it surfaces the same gaps, the kit closes them.
Get the same fixes as a do-it-yourself kit
The $29 ChatGPT-Visibility Kit: 18 audit prompts, copy-paste schema & FAQ templates (the exact ones we used here), a 22-point consistency checklist, and a tracker.
Get the ChatGPT-Visibility Kit — $29 Or have us set it up →Original, AI-assisted work. No ranking/placement guarantees — the AI platforms control their outputs; we get the inputs in your control done right. Not legal/financial advice.