Peec AI API alternative

Choose the right fit

From collection to a reviewable report

1. 01

   Collect answers

2. 02

   Save dated evidence

3. 03

   Analyze the sample

4. 04

   Deliver the report

The included Node runner checks live prices and a run budget, saves answer evidence to local JSON/CSV, resumes pending cells, and compares paired successful samples. Reuse saved data for reports. Your scheduler starts future runs; a crashed or timed-out request stays unknown until reviewed.

Available API building blocks

Which engines and regions can you query?

Live custom prompts use Google AI Mode via /geo/serp, or ChatGPT and Gemini web answers via /geo/answers. Indexed mention search and aggregation remain separate: engine=google covers Google AI Overviews; engine=chatgpt covers ChatGPT in the United States, in English. Perplexity and Claude live prompts are not supported. UI translations do not expand data coverage.

Make your first AI visibility request

curl --max-time 150 --request POST 'https://api.unifapi.com/api/geo/answers' \
  --header 'Authorization: Bearer YOUR_UNIFAPI_KEY' \
  --header 'X-Unifapi-Max-Credits: 30' \
  --header 'Content-Type: application/json' \
  --data '{
  "prompt": "Which project management software works well for a small team?",
  "engine": "chatgpt",
  "location": "us",
  "language": "en",
  "brands": [
    {
      "id": "your-brand",
      "names": [
        "Example Brand"
      ],
      "domains": [
        "example.com"
      ]
    }
  ]
}'

curl --max-time 150 --request POST 'https://api.unifapi.com/api/geo/mentions/search' \
  --header 'Authorization: Bearer YOUR_UNIFAPI_KEY' \
  --header 'Content-Type: application/json' \
  --data '{
  "target": [
    {
      "domain": "example.com"
    }
  ],
  "engine": "google",
  "language": "en",
  "limit": 20,
  "view": "standard"
}'

Illustrative requests, not measured results. Replace the prompt, aliases and domains. Live answers can take up to 120 seconds; the example waits 150 seconds and caps that request at 30 credits. Inspect credits_charged. A valid no-answer response is charged; collection errors are not. Do not automatically replay a timeout with an unknown outcome.

Move the workflow, keep the evidence

Start with a small set of queries and domains. Match language, location, engine, date, and sampling method before comparing results. Update authentication, request bodies, and response handling: this is not a drop-in implementation of Peec AI’s API and does not export its private project history.

Compare measurements, not just scores

Mention coverage divides successful samples mentioning a brand by all successful samples. Tracked-brand mention share instead divides that brand’s mention samples by the sum of mention samples for every tracked brand. Brands can overlap, so these denominators differ. Count citations separately. Valid no-answer samples remain in coverage; failed or unknown collections do not. For a trend, compare paired successful cells from the same panel and settings. Neither metric measures visits or conversions.

What to check in a trial or demo

For an agency, test whether you need a shared client workspace with daily prompt history or only evidence at report time. Ask Peec which API objects your plan exposes, whether historical answers are exportable, and which models and countries are included. With UnifAPI, define your own client IDs and retain dated answer snapshots.

Plan for the differences

The included runner can repeat your own prompt panel, but UnifAPI does not host its schedule or a project dashboard. Monthly indexed history is available; it is not a replay of your earlier custom prompts. First-party bot logs, sentiment or recommendation-rank scoring, Perplexity/Claude live prompts, and automatic website changes are not included.

Peec’s tracked-prompt history and visibility scores will not transfer automatically. Re-establish your baseline using the same queries, sampling window, and data source.

Questions before you switch