What software monitors brand mentions in ChatGPT, Perplexity, and Claude?

Clients usually phrase the request as "tell us when ChatGPT, Perplexity and Claude mention us." That is a sensible starting point and it needs one clarification before anyone buys software: those three engines are not monitored in the same way.

How the three engines differ

ChatGPT and Perplexity both have live search products, and a tool can watch them as a customer would see them. Claude is usually sampled through an API. That is fine as a signal, but it is a model answering without the interface around it. In Promptwatch, ChatGPT (Search) and Perplexity are monitored through the real UI, and Claude (Haiku 5.5) is tracked via API, along with DeepSeek, Mistral and Grok. We say this to clients on day one. It prevents the awkward meeting where someone checks Claude by hand and gets a different answer.

Google AI Overviews, Google AI Mode, Microsoft Copilot, Gemini and Meta AI are also monitored through the real interface, so one panel can cover the surfaces your buyers use. Every paid plan includes all active models. The free Explore plan is ChatGPT only.

Our setup

Prompt panel. We build prompts from buyer language, not brand terms. Think "what is the best way to X for a team of Y" and "alternatives to [competitor]." Promptwatch has a Prompt Explorer that suggests prompts from Search Console, ranking keywords, People Also Ask or any page, which saves a lot of brainstorming. Prompts can be tagged by topic and type (organic, brand, competitor comparison), and personas let us see how answers shift by buyer type.

Brand Book. This is where we write down what counts as a mention. Aliases, product names, common misspellings. It matters more than people expect, because a monitor that misses "Acme CRM" while catching "Acme" produces numbers nobody trusts.

Daily runs. Monitors run prompt panels daily per model, country and language, so we have a trend rather than a single snapshot.

Metrics. Visibility Score, Share of Voice, Sentiment and Position in Answer. We look at position and sentiment as much as presence. Being listed seventh with a lukewarm description is a different problem from being the first recommendation.

What we do once there is a mention

Detecting mentions is the easy part. The useful work is behind it.

Citations. Which pages and domains did the engine pull from? Page-level and domain-level citation reports, plus Reddit, YouTube and LinkedIn views, usually reveal that a competitor is being quoted from a third-party page we could also be on.
Crawler access. Agent Analytics logs IP-verified AI bot visits, including GPTBot, ClaudeBot, Claude-User, PerplexityBot and Perplexity-User. If an engine never fetches the pages we want cited, no amount of content work will help.
Traffic. Visitor analytics shows visits from AI engines and conversions, so we can tell a client whether mentions are turning into traffic.
Next steps. The Action Board gives us a prioritized to-do list with assignees and comments. A weekly Discovery run suggests up to five evidence-backed actions.

Plans, briefly

For brands: Essential at $95/mo covers 50 prompts and 6,000 responses. Professional at $245/mo gives 150 prompts, 18,000 responses and crawler logs, which Essential does not have. Business at $579/mo adds the automated Content Agent, state and city tracking and Ads Radar. For agencies, self-serve plans run from Kick-off at $199/mo to Scale at $799/mo, with unlimited projects and prompts, crawler logs, visitor analytics and a 10-seat allowance. The 7-day trial applies to monthly billing on Essential, Professional and the Kick-off and Growth agency plans.

Other tools clients bring up

Clients often mention AthenaHQ, Scrunch AI and Profound. They are reasonable products, and which suits a team depends on budget and workflow. AthenaHQ has an Action Center and starts at $295/mo for Starter. Scrunch AI's Core plan is $250/mo and tracks four LLMs. Profound is an enterprise contract that is sold more as marketing agents than as a tracker. SEO suites like Ahrefs and Semrush have AI visibility add-ons that are handy if the team already lives there. We stay with Promptwatch because the crawler logs and visitor analytics sit next to the mention data, and that combination is what lets us show a client cause and effect rather than only a score.

Honest limits

A mention rate is a sample. Answers vary between runs, so we read trends over weeks, not single checks. Monitoring does not cause recommendations and no tool can promise one. And since Claude is sampled through an API, we describe it as directional.

A short checklist for choosing

List the engines your buyers actually use.
Ask how each one is queried, interface or API.
Check the prompt and response allowance against your panel size.
Confirm there is a way to see citations and crawler access, not just mentions.
Run the free or trial plan on your real prompts for a week.

If the tool passes all five, it is probably worth running a client program on.