Bing Copilot AI Answers API | Microsoft AEO Monitoring

Track what Microsoft's answer engine says about any topic - and whether it cites YOUR brand - as clean, structured JSON.

Bing Copilot is the AI answer layer of Microsoft's search stack: for a growing share of queries, users read its generated answer instead of clicking ten blue links. If you do answer-engine optimization (AEO), you need to know what that answer says, which sources it cites, and whether your brand is in it. This API resolves any query to its Copilot answer: the headline summary, the full structured answer (headings, lists, tables, code), every cited reference, and an optional brand-mention check.

This is the fourth engine in the AEO suite, next to Google AI Overview, Naver AI Overview, and Brave AI Mode. Same input shape, same output style: monitor all four answer engines with one pipeline.

What you get for every query

Field	Description
Answer header	The one-paragraph headline answer
Markdown	The full answer rendered as a single markdown document, ready for diffing and LLM pipelines
Text blocks	The raw structured blocks: paragraphs, headings, nested lists, tables, code blocks
References	Every cited source with title, link, snippet, and source name
Brand check	brandMentioned boolean when you set brandToTrack
Extras	Featured video, image and video result links when present
Summary	A one-line human-readable summary built for AI agents

Use cases

• AEO / GEO brand monitoring - is your brand cited when Copilot answers "best crm for startups"? Track it weekly across your whole keyword set.
• Citation-source analysis - which domains does Microsoft's answer engine trust for your category? Target those for PR and content placement.
• Competitor share-of-voice - run the same query list with each competitor as brandToTrack and compare mention rates.
• Content strategy - see the exact structure (headings, tables, steps) the answer engine prefers for your keywords, and shape your content to match.
• Answer drift tracking - store the markdown over time and diff how the AI answer changes after your content updates.

Input

{
  "queries": ["best crm for startups", "best help desk software"],
  "brandToTrack": "Salesforce"
}

Parameter	Type	Notes
query / queries	string / array	The question(s) to resolve. At least one required.
brandToTrack	string	Optional. Adds a brandMentioned boolean to every answer row.

Answers typically take about 10 seconds per query to generate upstream - that is the answer engine thinking, not the actor being slow.

Example output

{
  "result_type": "copilot_answer",
  "query": "best crm for startups",
  "answerPresent": true,
  "answerHeader": "The best CRM for startups depends on budget and team size, with HubSpot, Pipedrive, and Salesforce among the most recommended.",
  "markdown": "## Top CRM picks\n\n- HubSpot: generous free tier\n- Pipedrive: pipeline-first UX\n- Salesforce: scales with you",
  "references": [
    { "index": 0, "title": "Best CRMs for Startups (2026)", "link": "https://example.com/best-crms", "source": "Example" }
  ],
  "brandToTrack": "Salesforce",
  "brandMentioned": true,
  "summary": "The best CRM for startups depends on budget and team size, with HubSpot, Pipedrive, and Salesforce among the most recommended.",
  "fetched_at": "2026-07-06T15:37:00+00:00"
}

Pricing

Pay-per-event: one flat charge per query resolved, whether you run one query or a thousand. No setup fee, no per-row surprises: one query, one answer, one charge.

How to get started

1. View on Apify Store and click Try for free.
2. Enter a query (and optionally the brand you want to track).
3. Run it and read the answer, citations, and brand check as JSON, CSV, or via the API.

Monitoring the full AEO landscape? Pair this with Google AI Overview, Naver AI Overview, and Brave AI Mode.

Prefer to start from code? The example repo on GitHub has a Python quick-start plus MCP install guides for Claude, Cursor, and ChatGPT.

🔌 Use this API from Claude (MCP)

Add this actor as a tool in Claude Code (free trial), Claude Cowork (free trial), or any other MCP client, via the hosted Apify MCP server. Use this actor-specific URL:

https://mcp.apify.com/?tools=actors,docs,johnvc/bing-copilot-api

Setup walkthrough:

Apify MCP integration docs: https://docs.apify.com/platform/integrations/mcp

Then ask things like: "Check whether Acme is mentioned when Microsoft's AI answers 'best project management software', and list the cited sources."

🔌 Integrations: Automate AEO Tracking and AI Citation Monitoring

A single run is a snapshot of what Microsoft's answer engine says today. The value compounds when you run it on a schedule and watch the answer, its cited sources, and your brand's mentions shift week over week. The Apify platform turns this API into an AEO tracking and AI citation monitoring pipeline. See all Apify platform integrations.

Scheduled AEO tracking with Tasks

Save one task per keyword set you monitor (for example your category's buying-intent questions), then attach a single schedule that runs them all. From the Actor, open Actions, then Schedule. Useful cron values: 0 6 * * 1 for a Monday-morning pull, 0 7 1 * * for the first of each month, 0 */12 * * * twice a day. One schedule can track a whole keyword set, so your brand-mention and citation data stay current.

n8n, Make, and Zapier

Route each new answer or brand-mention change straight into Slack, a sheet, or your dashboard. n8n, Make, and Zapier follow the same pattern: schedule trigger, run the Actor, filter for a brandMentioned flip or a new cited source, then notify or store. In n8n, install the dedicated community node n8n-nodes-bing-copilot-api (Settings, Community Nodes) to run this API in any workflow; it also works as an AI Agent tool.

Store answer history in Supabase for answer-drift tracking

Keep every run so you can diff how the AI answer changes after your content updates. The no-code path is the n8n Actor node into a Supabase node. In code:

from apify_client import ApifyClient
from supabase import create_client

apify = ApifyClient("<APIFY_TOKEN>")
supabase = create_client("<SUPABASE_URL>", "<SUPABASE_KEY>")

run = apify.actor("johnvc/bing-copilot-api").call(
    run_input={"queries": ["best crm for startups"], "brandToTrack": "Acme"}
)

rows = []
for a in apify.dataset(run["defaultDatasetId"]).iterate_items():
    rows.append({
        "query": a.get("query"),
        "answer_header": a.get("answerHeader"),
        "markdown": a.get("markdown"),
        "brand_mentioned": a.get("brandMentioned"),
        "references": a.get("references"),
        "fetched_at": a.get("fetched_at"),
    })

supabase.table("aeo_answers").insert(rows).execute()

Use it from an AI agent (MCP)

The dedicated MCP section above shows how to add this API to Claude, Cursor, or any MCP client. An agent can then answer "did our brand get cited when Copilot answered our top ten buying-intent questions this week?" and write the results to a sheet.

Webhooks

Add an Apify webhook on ACTOR.RUN.SUCCEEDED to push each finished run into your own AEO dashboard or alerting system.

🔗 Related Tools

Monitor every answer engine with one pipeline. This is the Bing Copilot engine in a four-part AEO suite:

• Google AI Overview API - the AI answer at the top of Google Search, the largest answer surface.
• Naver AI Overview API - AEO monitoring for Naver, the dominant search engine in South Korea.
• Brave AI Mode API - the AI answer from Brave's privacy-first search.

Comparing Bing Copilot tools? An alternative such as Bing AI Search (Copilot) Answers returns the Copilot answer but has thin adoption, no brand-mention check, and no AEO-suite companions for the other engines. This API adds the brandMentioned check, the structured citation list, and three sibling engines so you monitor Google, Naver, Brave, and Bing from one place.

FAQ

Why does a query take around 10 seconds?

The answer engine generates its response live, which typically takes 8 to 13 seconds upstream. Batch your queries into one run with the queries list; they are processed sequentially with one charge each.

What if there is no AI answer for my query?

You still get a row with answerPresent: false and a note, so your monitoring pipeline can tell "no answer" apart from "run failed". The query is still charged, since the engine was queried.

How is brandMentioned computed?

Case-insensitive match of your brandToTrack string across the full answer text (all blocks) and the cited references (titles, links, snippets, and source names).

Can I set the country or language?

Not yet: the upstream answer engine currently exposes no country or language controls for this answer surface. If Microsoft adds them, they will be added here.

What is answer engine optimization (AEO)?

Answer engine optimization is the practice of shaping your content so AI answer engines cite and recommend you, the way search engine optimization targets the ten blue links. As people read AI answers instead of clicking through, being named in the answer matters more than ranking below it. This API is the measurement half of AEO tracking for Microsoft's Bing Copilot: it shows what the answer says, which sources it cites, and whether your brand is in it.

Should I use an API or a web scraper for AI answers?

Use the API. A brittle web scraper breaks whenever the answer surface changes and needs proxies and browser automation. This is a maintained API: send a query, get back the answer, its structured blocks, the cited references, and a brand-mention check as clean JSON, and let the Apify platform handle scale and reliability.

Can I track my brand across ChatGPT and other AI engines?

This API covers the Bing Copilot engine. For full coverage, pair it with the three sibling engines in the Related Tools section: Google AI Overview, Naver AI Overview, and Brave AI Mode. Run the same query list through each with your brand as brandToTrack for a cross-engine share-of-voice view. All four share the same input and output shape, so one pipeline drives AI citation tracking across every answer surface.

Can I use Bing Copilot data through an MCP server?

Yes. Add the Actor to any MCP client (Claude, Cursor, and others) through the hosted Apify MCP server using the URL in the MCP section above. It works in Claude Code (free trial) and Claude Cowork (free trial).

Can I use this with the Apify API?

Yes. The Apify API runs the Actor, schedules it, and returns datasets programmatically. Use the apify-client package for Python or Node.js, or call the REST endpoint directly. See the Integrations section for a scheduled AEO tracking recipe.

💡 Example tasks

Ready-to-run configurations for common use cases. Open one, change the inputs, and run:

• AEO Tracking: Your Brand in Bing Copilot Answers
• AI Citation Tracking for Bing Copilot Answers
• Bing Copilot API: AI Answers and Citations as JSON
• Monitor Your Brand in Microsoft Copilot Answers
• Track Your Brand Across AI Answer Engines

Where the answers come from

Every row in the dataset is the public AI answer that Microsoft Copilot returns for your query, captured as it was served: the headline answerHeader, the full answer as markdown, the structured textBlocks (paragraphs, headings, nested lists, tables, code), and the references list carrying the index, title, link, snippet, and source name of each citation. Nothing is rewritten or summarized on top of it. When the engine has no answer for a query, you get a row with answerPresent: false and a note rather than a silent gap in your data.

This API is an independent tool published on Apify. It is not affiliated with, endorsed by, or operated by Microsoft, and it is not Microsoft's official API. It reads the same public answer surface you can read yourself and hands it back as structured data.

How is this different from asking Bing Copilot by hand?

By hand, a single keyword means opening Bing Copilot, waiting out the eight to thirteen seconds the answer takes to generate, reading it, then clicking through each citation to note the domain. Repeat that across forty buying-intent keywords every Monday and it eats a morning, and you still end up with notes instead of a diffable record. Here you pass the keywords as queries, set brandToTrack, and get one JSON row per query holding the answer, its cited sources, and a brandMentioned boolean you can store and compare week over week.

Does Microsoft Copilot have a public API for its answers?

Not one that hands you the answer surface as structured data. Microsoft Copilot is built to be read in a browser, so there is no documented endpoint that returns the answer blocks, the cited references, and a brand check in a shape you can schedule, diff, and pipe into a dashboard. That is the gap this API closes: the same public answer, delivered as JSON, one charge per query resolved.

Last Updated: 2026.07.27