Everything AI says about your market, and what to do next
Toucan Intelligence asks the questions your customers ask, reads every answer, and turns the citations and recommendations inside them into reports and plans.
Discover
The data layer. Same for every account.
ChatGPT, Gemini, Claude and Perplexity, queried independently. Each cites different sources; you see all of them.
Every publication, URL and source reference in every answer, with the brand it supports and the model that used it.
Being named in a list is not a recommendation. We separate mentions, picks and first picks, and record the reason given.
AI cites different sources in different markets. Run the same queries in the US, UK, DE and 12 more.
News, Reddit, YouTube, affiliates, official pages. See which source types carry weight for your prompts.
Every run is kept. Compare share, citations and pages between runs to see what your work changed.
Reports that lead with your goal
Choose brand or publisher at sign-up. Switch any time in Settings.
- Recommendation rate and share of recommendations, ranked against competitors
- Recommendations by query: first pick, named, mentioned, absent, for every model and region
- Why AI recommends: the reasons attached to your picks vs the leader’s
- Sources shaping recommendations, ranked for outreach
- Citation share against every source type, and your rank among affiliates
- Most cited pages by model, with gains and losses between runs
- Prompt coverage and the biggest gaps, with who is cited instead
- Which brands AI recommends using your pages as evidence
NewAct
Three tools that turn a run into work. Each ends with an optional hand-off to the Toucan team.
Brief it with a budget band, tone, channels you can’t use and regions to prioritise. It drafts up to five plays from the run’s citation gaps, each linked to its evidence. Track status and owner per play, pin what’s working, exclude what isn’t, and regenerate the rest. Every version shows what changed and why.
Pick your page and a prompt. We fetch the pages AI cites for that prompt and score everything on 16 checks across authorship, freshness, trust, on-page technical and depth, using Google’s Quality Rater Guidelines as the reference. You get a ranked fix list with effort and estimated gain.
AI crawlers fetch raw HTML and mostly don’t run JavaScript. The crawler view measures what they get: time to first byte with bot user-agents, how much of your content exists without JS, and whether GPTBot, ClaudeBot and PerplexityBot are allowed in. Lighthouse Core Web Vitals sit alongside for the visitor side.