Hiveminds Lens analyses how visible your business is in ChatGPT, Claude, Perplexity and Google AI Overviews. We test relevant prompts, compare you to competitors and show which citation, technical and content signals are missing.
Lens is not a one-off report but a loop that keeps repeating: observe what AI systems say about your brand, diagnose where you fall short, intervene with targeted fixes, re-measure on an identical yardstick and keep adjusting. Every step below describes what is actually in the product.
Everything we track and how we score is public in our Lens methodology (Dutch)What do AI systems say about your brand?
Every improvement starts by looking. We record what ChatGPT, Claude, Perplexity and Google AI Overviews say about your brand on questions your customers actually ask, what sources they draw on and how you compare to competitors. We also study your server logs: what AI bots visit your site, how often, and on what pages.
View live bot data at /lens/liveEnter a URL and see what HTML a crawler receives from your site, and how three OpenAI agents are treated in your robots.txt: GPTBot (training crawler), OAI-SearchBot (drives ChatGPT Search inclusion) and ChatGPT-User (retrieves pages when a user asks). Three agents, three distinct roles: GPTBot being able to read your site does not make you findable in ChatGPT. No email required.
Check your own site
Free, no email required.
Want a full analysis of where you stand in ChatGPT, Claude, Perplexity and Google AI Overviews? Request a free Lens Snapshot.
This is what an AI search query looks like, and why your competitors get mentioned.
Demo analysis: example prompts| Prompt | Status | Platform |
|---|---|---|
| ✓ Mentioned | ChatGPT | |
| ✗ Absent | Perplexity | |
| ✓ Mentioned | Google AI Overviews | |
| ✗ Absent | ChatGPT | |
| ✓ Mentioned | Perplexity |
How your AI visibility compares to the key players in your market.
Example: mortgage market| Your business | Competitor A | Competitor B | Competitor C | |
|---|---|---|---|---|
| ChatGPT mentions | 2 | 18 | 24 | 11 |
| Perplexity mentions | 1 | 14 | 19 | 9 |
| AI Overviews presence | 0% | 60% | 75% | 35% |
| Technical readiness | 42/100 | 78/100 | 84/100 | 65/100 |
| Content depth (pages) | 23 | 187 | 234 | 78 |
Where do you fall short: technology, content or citations?
If AI systems rarely name you, we want to know exactly why. Lens tracks three dimensions: Technical (can AI crawlers fetch and process your site), Content (is your content structured so models can read and reuse it) and Citation (do all four platforms actually draw on you as a source).
Together, those three dimensions form your Lens score: a weighted average in which Citation weighs heaviest (0.35), followed by Technical (0.30) and Content (0.20). Weights are normalised and you get a score from 0 to 100. One number tells you where you stand; each dimension tells you where to focus.
Read how every score is built at /lens/methodologieAI systems go through three stages before answering. They fetch sources (crawl and index), they read and understand content (context analysis), and they choose what sources to cite. Those three stages map exactly onto our three dimensions: fetch is Technical, read is Content, choose is Citation.
Crawl and index. What AI bots visit your site, how often, what pages?
Understand and classify. How do models interpret your content?
Cite and recommend. When and where does your brand appear in AI answers?
What prevents AI search engines from finding and citing your content?
Add structured data with JSON-LD for your business information
universalRemove the Disallow rule for OAI-SearchBot in robots.txt
universalImplement server-side rendering or pre-rendering for crucial content
B2B SaaSPublish comparison content that AI models can cite as a source
e-commerceBuild external mentions as entity anchors for LLM recognition
consultingCan help crawlers, no proven ranking impact, but low effort
optionalSitemap correctly configured and findable
universal| Crawler | Status | Access |
|---|---|---|
| GPTBot | Verified | Blocked |
| OAI-SearchBot | Verified | Blocked |
| ChatGPT-User | Verified | Allowed |
| ClaudeBot | Verified | Allowed |
| Claude-Web | Verified | Allowed |
| PerplexityBot | Verified | Allowed |
| Google-Extended | Verified | Blocked |
Based on robots.txt analysis: 3 out of 7 AI crawlers blocked
For a decade, more content, more backlinks and higher rankings was a winning formula. AI search engines work fundamentally differently. ChatGPT, Claude, Perplexity and Google AI Overviews generate answers from sources they treat as authoritative. They cite, they compare, they conclude. Being named in a 1-on-1 conversation with a buyer who has purchase intent beats a thousand generic clicks.
What fix do we apply, and when?
Each diagnosis leads to a concrete intervention: a technical fix (robots.txt or rendering, say), a content change, or work on sources that earn you citations. We log every intervention as a fix event: what we did, when, on what pages. Every later score change hangs on a verifiable moment, so you can always trace what action had what effect.
Prompt Path Mapping charts how AI systems reason from first question to vendor choice. It is what Lens rests on. We analyse where your brand is absent, what sources models cite, and what signals you need to appear as a logical conclusion. That is how we decide what fix takes priority.
Read our full definition at /prompt-path-mappingWhat changed after each fix?
After a fix we re-run on exactly that same pinned prompt and platform set as before: identical prompt texts in identical order, on all four platforms. Apples to apples, so re-runs reveal per dimension what changed. We do at most one targeted re-run per fix event, so every result stays tied to one intervention.
A delta after an intervention is an observation, not causal proof. AI answers vary per run and per model update. How we deal with that, and why a pinned set is indispensable, is covered in our methodology.
How re-runs and pinned sets workMeasure weekly, keep adjusting
AI answer presence is never a one-off project. On a retainer we run one weekly check per domain, at a fixed slot in Amsterdam time, always on your pinned prompt and platform set. A dashboard shows your latest numbers and their course over time, and in a monthly session we discuss results and your next best action. In short: weekly runs, monthly reviews.
You can ask a prompt yourself and check if your business gets named. That is anecdotal. AI answers change per session, per model update, per region. Lens runs systematically: identical prompts on four AI platforms, every week, on a pinned prompt and platform set.
We calculate one score from three tracked dimensions: Citation (0.35), Technical (0.30) and Content (0.20), normalised to a 0 to 100 scale. Every run stores what scoring version produced it, so you can compare scores over time honestly.
A loose prompt gives you a feeling. Lens gives you a yardstick.
This is what the Lens dashboard looks like for clients on an Improve retainer. Weekly tracking, prompt intelligence and crawler analysis.
Lens works in three layers: a one-off Snapshot as baseline, weekly monitoring on your pinned prompt and platform set on a retainer, and targeted re-runs around fix events. Our products follow those layers.
A one-off baseline of your AI presence. We run relevant prompts on ChatGPT, Claude, Perplexity and Google AI Overviews, compare you to competitors and flag technical blockers. Report in your inbox. No cost, no obligations.
In-depth, partly manual prompt analysis on four AI platforms: ChatGPT, Claude, Perplexity and Google AI Overviews. Scoring per our public methodology plus server log analysis. Full report with concrete recommendations and a walkthrough session.
Ongoing AI presence work with interpretive consultancy. A dashboard on weekly runs, Prompt Path Mapping as baseline, targeted re-runs around fix events, monthly reporting and review sessions, and content plus distribution strategy.
Receive an initial analysis of your AI visibility within 5 minutes. No commitment, no cost.