AI engines

AmICited asks each of your prompts to eight AI engines and records what every one of them answered: the text, the brands it named and the sources it cited. Five engines are included on every plan. Three more (Claude, Grok and DeepSeek) are premium engines available on the Premium and Enterprise plans.

The eight engines#

Where an engine lets AmICited choose, web search is switched on (ChatGPT’s is forced on, and Claude, Grok and DeepSeek each use their own search tool), because that is how people use these engines to research purchases, and it is what makes citations possible. Each engine run produces one AI response, and each response is metered in credits (see cost per engine below).

How the engines differ#

Engines disagree often, which is why AmICited tracks them separately and lets you filter every chart by engine.

• ChatGPT and Gemini answer in a chat format and often name several brands in a list. They also report the sub-queries they searched, which feed query fan-out directly.
• Perplexity is collected from a real browser session in the prompt’s country, so you see the answer a local user would see, including its cited sources.
• Google AI Overview only appears when Google decides to show one for that query. Prompts phrased like searches are more likely to trigger it than long conversational ones.
• Google AI Mode is a separate surface from AI Overview. The two cite different URLs most of the time, so do not treat one as a proxy for the other.
• Claude, Grok and DeepSeek each run their own web search rather than borrowing Google’s, so they often surface a different set of sources again.

Premium engines#

Claude, Grok and DeepSeek are queried through the vendors’ own paid APIs rather than a scraping service. That is why they are reserved for the Premium and Enterprise plans. On lower plans they appear in the provider picker as locked, with an upgrade prompt, and cannot be selected.

Choose engines per prompt#

Every prompt carries an optional engine allow-list, shown as AI providers in the setup wizard, the Add prompts dialog and Edit prompt.

• All selected (the default). The prompt runs on every engine your plan includes. When AmICited adds a new engine, these prompts pick it up automatically.
• A subset. Untick engines to restrict the prompt. Only the ticked engines run, and new engines are not added. You must keep at least one.

Use a subset to save credits on prompts where some engines do not matter to you, for example restricting local-service prompts to Google AI Overview and Google AI Mode. Keep the default for your core category prompts so the comparison across engines stays complete.

Filter results by engine#

The engine filter at the top of the Dashboard, Prompts, Sources and Competitors pages narrows every figure to one engine. The By platform breakdown on the dashboard shows your visibility per engine side by side. When your blended visibility score moves, filter by engine first: most swings come from one engine changing how it answers.

Cost per engine#

Each engine run is one AI response:

• Standard engines (ChatGPT, Perplexity, Gemini, Google AI Overview, Google AI Mode) cost a flat 0.015 credits per response.
• Premium engines (Claude, Grok, DeepSeek) are metered at the vendor’s actual cost for the tokens and web searches used, and never less than 0.015 credits per response.

So one prompt run once on the five standard engines costs 5 × 0.015 = 0.075 credits. See run frequency for monthly budgets and credits for everything else that is metered.

Related#