Thinking.Whaily tracks how ChatGPT, Claude, Gemini, Perplexity and DeepSeek answer your buyers' questions, names the exact model version behind every number, and shows the sources that shape the answers. Numbers a marketer can defend, at a price a small team can pay.
Start freeChatGPT
Claude
Gemini
Perplexity
DeepSeek
GrokWhat you get
These are the console's own components, rendered here with example data. Click a model in the legend and the chart responds, because it is the same chart.
Live in the product
43.7%
Visibility
Named or recommended in 561 of 1284 responses
18.2%
Share of voice
704 of 3868 times a tracked brand was named
51.4%
Inclusion rate
Of the answers naming any tracked brand, the share that named you
#3.4
Average position
First mention, lower is better
1284
Responses
Tracked prompt runs in this window
Example data, not a customer's. Every number in the product names the model version it came from.
Neural Cluster Influence
AI answers are built from a small set of third-party pages: top lists, review sites, forums, documentation. Whaily records every page the models cite in your category and scores each one by how much weight the answers give it.
That's the NCI score.
Everything you need
Visibility, share of voice and position across Claude, GPT, Gemini, Perplexity and DeepSeek, with a confidence interval on every number and the model version named on every row.
Learn more about Know where you stand, by exact model versionEvery top list, review site, forum and article the models cited in your category, with who was named alongside, scored by influence so you know which pages are worth winning.
Learn more about Find the pages AI actually citesSentiment from positive to negative and position from first to last, for you and every tracked competitor, on the same answers, so the comparison is fair.
Learn more about See how each model describes you, prompt by promptThirty-four checks across Wikipedia and knowledge sources, review sites, social profiles and the technical basics, each with a status and one line that explains it.
Learn more about Audit your presence where AI looksEighteen rules read your answers, sources and audit every night and produce ranked recommendations. Convert one to a task, assign it, and see whether it worked.
Learn more about Turn gaps into a prioritised action planHead-to-head: the prompts where a competitor was named and you were not, the sources that back them and not you, and who is moving up this week.
Learn more about See exactly where competitors beat youHow we measure
AI answers change from one run to the next. Most tools hide that. Whaily shows the model, the sample and the range behind every number, and tells you what it does not measure.
Not "ChatGPT" but gpt-5.4, not "Claude" but claude-sonnet-4-6. Pin a version or follow the latest release, and get an email before a provider retires the one you track. A chart move after a version change is marked on the timeline, so it is never a mystery.
Models and versionsOne AI answer is one sample of a system that varies run to run. Share of voice carries a 95 percent confidence interval, a metric with fewer than five answers shows N/A instead of a fake 100 percent, and a burst of repeated runs turns a guess into a measurement when a decision needs one.
Bursts and confidenceWhaily queries each model through its API, which makes results comparable across versions and days. It also means we do not measure the ChatGPT web interface, Google AI Mode, Copilot or Meta AI, where a router, a system prompt and memory change what a logged-in user sees. We say that out loud.
How we publishEvery computed value stores the version of the method that produced it. When the method changes, the history is backfilled on purpose, never rewritten quietly. Burst runs are kept out of the daily tracking numbers so a big experiment cannot bend a trend line.
What we verify, and whenSources like top-lists, forums and review platforms carry an NCI score. That score reflects how much weight AI places on them when forming answers in your category.
| Source | NCI | A | B | C |
|---|---|---|---|---|
G2 | 0 | · | ||
TechCrunch | 0 | · | ||
Reddit community | 0 | |||
Top 10 SaaS list | 0 | · | · | |
Expert roundup | 0 | · | · |
| Model | Comp A | Comp B | Comp C |
|---|---|---|---|
| ChatGPT | 73% | 41% | 58% |
| Gemini | 65% | 53% | 49% |
| Claude | 81% | 38% | 62% |
| Perplexity | 69% | 47% | 55% |
| DeepSeek | 44% | 31% | 39% |
| Grok | 52% | 36% | 47% |
Sentiment analysis
Visibility tells you whether you appear. Sentiment and position tell you how: positive, neutral or negative, and named first or after three competitors. Whaily scores you and every tracked competitor on the same answers, with the criteria buyers use as the rows.
Public tiers, no contact-us gates
Quote-only, opaque enterprise pricing
Public tiers, hidden overage fees
Native to most stacks, light on niche tools
Broad catalogue, deep enterprise hooks
Limited to top vendors, weak on Salesforce
Minutes to first value, guided setup
Multi-week implementation, services-led
Self-serve but documentation gaps
Responsive, occasional Tier-1 friction
Dedicated CSM, strong on enterprise
Reports of slow responses, ticket-only
SOC 2, EU residency, clear DPA
SOC 2 + ISO 27001, full audit trail
SOC 2, residency limited to US
New · MCP server
Connect Claude, Cursor or any MCP client to your workspace. Ask about visibility, sources and competitors, get the same numbers as the dashboard, and act on them without leaving the conversation. The same keys open a REST API with 37 endpoints for the flows you build yourself.
Every number and list the dashboard shows is a tool your assistant can call: visibility, share of voice, competitors, sources, recommendations, audit checks, investigations. The analyze tool answers a custom question from the same query layer as the dashboard, so the two cannot disagree.
32 read tools
Create, tag, pause and import prompts, add competitors, convert a recommendation into a task, mark a domain as a partner, create a share link. Real changes in your workspace, from the same conversation, behind a key with the manage scope.
40 manage tools
Keys carry a scope (read, manage, expensive), never expire unless you ask, rotate with a grace window and revoke instantly. Every call is rate limited, capped per month by plan and written to an audit log. The organisation comes from the key, never from the request.
Scoped · audited · revocable
A selection of the 73 MCP tools
Pricing
Every paid plan runs on Whaily's AI provider keys by default, or bring your own for 40% off.
See where AI sends people in your category. No card required.
or $0/month on your own keys
Budget models across every provider, daily tracking.
or $29.40/month on your own keys
Every default model, full competitive intelligence.
or $119.40/month on your own keys
Agency prices per client brand, Enterprise by contract. See full pricing
Start on a free trial, no card needed. Enter your domain and Whaily proposes the prompts and competitors before anything runs.
Get started free →