# AnswerLine

> One API for the live answers users get from ChatGPT, returned as structured JSON with sources and citations. Coming soon: Perplexity, Gemini, Microsoft Copilot, Grok, Google AI Mode, Google Search, and Google News.

## Docs
- [Documentation](https://answerline.dev/docs)
- [Quickstart](https://answerline.dev/docs/quickstart): From an API key to a synchronous answer, an async task and a webhook, with curl, the TypeScript SDK and the Python SDK.
- [Authentication and errors](https://answerline.dev/docs/authentication): API keys, the error envelope and every error code, which requests are safe to retry, and request ids.
- [Rate limits and concurrency](https://answerline.dev/docs/rate-limits): The one-second rate window, concurrency slots per plan, queue capacity and the headers that report them.
- [Credits and billing](https://answerline.dev/docs/credits): How requests are priced, reserved and charged, how billing cycles reset, and what happens when credits run out.
- [Async tasks and batches](https://answerline.dev/docs/async): Queue single tasks or batches of up to 500 with priorities and idempotency keys, collect results and manage the queue.
- [Webhooks](https://answerline.dev/docs/webhooks): Receive task results by webhook: payload, signature verification, secret rotation, retries and delivery order.
- [MCP server](https://answerline.dev/docs/mcp): Connect AI agents and MCP clients to the hosted MCP server: endpoints, authentication, tools and costs.
- [Integrations](https://answerline.dev/docs/integrations): Call the API from n8n, Zapier and LangChain for Python and JavaScript.
- [API reference](https://answerline.dev/docs/reference): Every endpoint, parameter and response code, generated from the OpenAPI document.
- [OpenAPI document](https://answerline.dev/openapi.json)
- [Full docs text](https://answerline.dev/llms-full.txt)
- [Pricing](https://answerline.dev/pricing)
- [Changelog](https://answerline.dev/changelog)

## Engines
- [ChatGPT](https://answerline.dev/engines/chatgpt): The answer ChatGPT gives real users, with every cited source, the searches it ran, product cards and brand mentions.
- [Perplexity](https://answerline.dev/engines/perplexity) (coming soon): Perplexity's answer with its citations, related queries, the searches it ran, and any videos, images, places, hotels or shopping cards it showed.
- [Gemini](https://answerline.dev/engines/gemini) (coming soon): Gemini's grounded answer with its sources, inline citations and places.
- [Microsoft Copilot](https://answerline.dev/engines/copilot) (coming soon): Copilot's Bing-grounded answer with its sources, the searches it ran, shopping cards and map results.
- [Grok](https://answerline.dev/engines/grok) (coming soon): Grok's answer with its cited sources, including X posts, and the searches it ran.
- [Google AI Mode](https://answerline.dev/engines/ai-mode) (coming soon): Google's conversational AI Mode answer with citation pills, places, shopping cards, inline products, videos and ads.
- [Google Search](https://answerline.dev/engines/google-search) (coming soon): Structured Google results: organic listings, ads, People Also Ask, knowledge graph, local pack and the AI Overview.
- [Google News](https://answerline.dev/engines/google-news) (coming soon): Google News results for any query, country and interface language, up to ten pages deep.

## Integrations
- [TypeScript / JavaScript SDK](https://answerline.dev/integrations/typescript): The official npm SDK: typed monitor and task methods, batches, webhooks and error classes for Node and edge runtimes.
- [Python SDK](https://answerline.dev/integrations/python): The official PyPI SDK: monitor calls, async tasks and batches with the same response shapes as the REST API.
- [Hosted MCP Server](https://answerline.dev/integrations/mcp): Give Claude, Cursor or any MCP client search and AI-answer tools in one line — no code, no infrastructure.
- [n8n Community Node](https://answerline.dev/integrations/n8n): Drop AnswerLine into n8n workflows: run monitors, queue tasks and read credits as native nodes.
- [Zapier App](https://answerline.dev/integrations/zapier): Trigger Zaps from AI answers: Run Monitor for quick calls, Create Task with a webhook callback for long-running engines.
- [LangChain Tools](https://answerline.dev/integrations/langchain): Ready-made LangChain tools for JavaScript and Python: one monitor tool per engine, plus task and credit tools.

## Use cases
- [AI Visibility Tracking API](https://answerline.dev/use-cases/ai-visibility-tracking): Track how ChatGPT, Perplexity and Google's AI answers describe a brand: mentions, citations and the wording users actually see.
- [Rank Tracking API](https://answerline.dev/use-cases/rank-tracking): Google positions, SERP features and the AI Overview for any keyword as structured JSON — one call, no headless browser fleet.
- [Local Rank Tracking API](https://answerline.dev/use-cases/local-rank-tracking): City- and state-level Google SERPs and AI answers: the same query, re-run from any location, returned as parsed JSON.
- [Competitor Analysis API](https://answerline.dev/use-cases/competitor-analysis): Pull a competitor's organic positions, ads and AI-answer citations for any keyword set — the same SERP data they see, as JSON.
- [Web Search API for Agents and Apps](https://answerline.dev/use-cases/web-search-api): Ground agents, RAG pipelines and in-product search on Google's index: ranked results and AI answers as JSON, per call.
- [Keyword Research API](https://answerline.dev/use-cases/keyword-research): Mine People Also Ask, related searches and live SERP composition programmatically — keyword discovery from the result page itself.
- [News Monitoring API](https://answerline.dev/use-cases/news-monitoring): Track coverage of a brand, product or topic across Google News and AI answers — structured results instead of an inbox of alerts.
- [Adverse Media Screening API](https://answerline.dev/use-cases/adverse-media-screening): The open-web layer of AML and KYC screening: continuous news, search and AI-answer monitoring for counterparties, per call.
- [Brand Protection API](https://answerline.dev/use-cases/brand-protection): Detect counterfeit sellers, typosquat ads and impersonation across Google SERPs and AI answers — the abuse signals, as data.
- [Threat Intelligence API](https://answerline.dev/use-cases/threat-intelligence): Automate open-web recon: search operators and AI-answer checks for exposed assets, phishing pages and impersonation infrastructure.

## Compare
- [Build vs buy](https://answerline.dev/compare/build-vs-buy): The real cost of capturing ChatGPT, Perplexity and Google AI answers in-house — browser fleets, parser maintenance, geo egress — versus per-call pricing.
- [Provider APIs vs AnswerLine](https://answerline.dev/compare/provider-apis): The OpenAI or Gemini API returns a completion — not the answer the consumer app shows. What differs, and when each is the right call.
- [SerpApi vs AnswerLine](https://answerline.dev/compare/serpapi): Google Search, Google News and AI Overview coverage and price per search on SerpApi, next to this API's credit prices for the same requests.
- [AI Overview API alternative](https://answerline.dev/compare/ai-overview-api-alternative): Who extracts Google's AI Overview today, what each returns, and this API's AI Overview fields from its OpenAPI document.

## Endpoints
- POST https://api.answerline.dev/v1/monitor/chatgpt
- POST https://api.answerline.dev/v1/monitor/gemini
- POST https://api.answerline.dev/v1/monitor/grok
- POST https://api.answerline.dev/v1/monitor/google
- POST https://api.answerline.dev/v1/monitor/google/goto
- POST https://api.answerline.dev/v1/monitor/google/news
- POST https://api.answerline.dev/v1/monitor/copilot
- POST https://api.answerline.dev/v1/monitor/perplexity
- POST https://api.answerline.dev/v1/monitor/aimode
- GET https://api.answerline.dev/v1/countries
- POST https://api.answerline.dev/v1/async/task
- POST https://api.answerline.dev/v1/async/task/batch
- GET https://api.answerline.dev/v1/async/task/{taskId}
- GET https://api.answerline.dev/v1/async/status
- DELETE https://api.answerline.dev/v1/async/queue
- GET https://api.answerline.dev/v1/credits
- GET https://api.answerline.dev/v1/states

## Blog
- [AI answer tracking as an agency deliverable](https://answerline.dev/blog/ai-answer-tracking-for-agencies): Clients are asking "do we show up in ChatGPT?" — how agencies can answer that as a service line without building infrastructure.
- [Measure the traffic AI engines actually send you](https://answerline.dev/blog/ai-referral-traffic): AI referral traffic is real but noisy — partial attribution, dark social, no query data. How to measure it properly and connect it to answer monitoring.
- [How often do AI answers change? Setting a monitoring cadence](https://answerline.dev/blog/ai-monitoring-cadence): AI answers drift on their own schedule — model updates, fresh sources, prompt phrasing. How to pick a sampling cadence per use case without wasting credits.
- [Monitor what AI answers say about your competitors](https://answerline.dev/blog/competitor-ai-answer-monitoring): Competitive intelligence from answer data — who gets recommended, on which prompts, in which markets, and when it changes.
- [Build vs buy: AI visibility tracking infrastructure](https://answerline.dev/blog/ai-visibility-build-vs-buy): The real cost of collecting ChatGPT, Perplexity and Google AI answers yourself — sessions, parsing drift, geo coverage — vs paying per call.
- [AI share of voice: a measurement framework](https://answerline.dev/blog/ai-share-of-voice-framework): Concrete metrics for AI visibility — mention rate, citation share, and coverage — with formulas you can compute from monitor API responses.
- [Track the products ChatGPT recommends](https://answerline.dev/blog/chatgpt-shopping-cards): ChatGPT shows product cards inside shopping answers. How to read them as data — prices, sellers, ratings — and monitor which products get recommended.
- [Capture Google AI Mode answers with an API](https://answerline.dev/blog/google-ai-mode-api): AI Mode is Google's full conversational answer surface — distinct from AI Overview. How to get its answers, sources and product blocks as JSON.
- [Getting Google News data programmatically](https://answerline.dev/blog/google-news-api): Google News has no official API for results. How to get headlines, clusters and sources as JSON — per country, language and device — and what to use it for.
- [Query fan-out: the searches ChatGPT runs before it answers](https://answerline.dev/blog/chatgpt-query-fan-out): Capture the web searches behind ChatGPT's answers and turn them, across a prompt set, into a list of queries worth winning.
- [How AI engines choose which sources to cite](https://answerline.dev/blog/how-ai-engines-choose-citations): What actually drives a citation inside ChatGPT, Perplexity or Gemini answers — the retrieval layer, fan-out queries, and the page traits that get picked.
- [AI answers differ by location — here's how to see it](https://answerline.dev/blog/local-ai-answers-state-level): The same ChatGPT prompt returns different answers in Texas and California. Geo-targeted monitoring at country and US-state level, and why local brands need it.
- [Idempotency keys are your retry safety net](https://answerline.dev/blog/idempotency-keys-explained): Retries are mandatory for async pipelines — and dangerous without idempotency. How keys prevent duplicate tasks and double charges.
- [How to measure brand mentions in ChatGPT answers](https://answerline.dev/blog/measuring-brand-mentions-chatgpt): A repeatable method for brand mention rate, citation share and shopping-card presence in ChatGPT answers, with guidance on how many prompts to run.
- [Give a LangChain agent live AI answers as a tool](https://answerline.dev/blog/langchain-agent-live-answers): The LangChain integration turns every engine into a tool call — your agent can ask ChatGPT, search Google or check News mid-reasoning. Setup in Python and JS.
- [Monitor Microsoft Copilot answers with an API](https://answerline.dev/blog/monitor-copilot-api): Copilot's grounded answers include sources, shopping cards and rich media. How to get them as structured JSON, geo-targeted, sync or async.
- [Giving an AI agent live answer data with the MCP server](https://answerline.dev/blog/mcp-server-agents): Connect an MCP client to the hosted server, let an agent query AI assistants as tools, and keep cost and latency in check.
- [Mentions vs citations: the two numbers that matter in AI answers](https://answerline.dev/blog/mentions-vs-citations): Being named in an answer and being cited as a source are different events with different fixes. How to measure and improve each.
- [Monitor Google Gemini answers with an API](https://answerline.dev/blog/monitor-gemini-api): Gemini's grounded answers cite Google's index directly. Get its answer text, sources and search queries as JSON — per country, sync or in batches.
- [Monitoring Google AI Overviews for your keywords](https://answerline.dev/blog/monitor-google-ai-overviews): Request AI Overviews with Google results, then measure how often one appears, which sources it cites, and how that compares with organic rank.
- [Monitor Perplexity answers with an API](https://answerline.dev/blog/monitor-perplexity-api): Get Perplexity's answer text, citations, related queries and media cards as JSON — sync or async — with Python and TypeScript examples.
- [Scheduled AI-answer monitoring in n8n](https://answerline.dev/blog/n8n-ai-visibility-workflow): A no-code pipeline — scheduled prompts, async tasks, webhook results into a sheet or database — built with the AnswerLine n8n node.
- [Monitor Grok answers with an API](https://answerline.dev/blog/monitor-grok-api): Grok pulls from X and the live web — a different evidence base than every other engine. How to capture its answers, citations and post sources as JSON.
- [Why AI engines cite Reddit so much — and what to do about it](https://answerline.dev/blog/reddit-citations-ai-answers): Reddit threads appear disproportionately in AI citations. The structural reasons, how to measure it for your prompts, and how to earn the citation anyway.
- [Building a rank-tracking pipeline for AI answers with batches](https://answerline.dev/blog/rank-tracking-pipeline-batches): Submit a prompt set as batches, collect results by webhook, and turn cited sources into a rank per prompt, market and day.
- [Mine People Also Ask for keyword research that still works](https://answerline.dev/blog/people-also-ask-keyword-research): PAA boxes are Google's own map of what people ask next. How to extract them at scale and turn them into a content plan.
- [Turn Google results into typed data, not HTML scraping](https://answerline.dev/blog/structured-serp-data-api): Organic results, ads, People Also Ask, local pack, AI Overview — one Google Search call returns every block as typed JSON, geo-targeted per country and ZIP.
- [Sync calls, async tasks or webhooks: how to call an answer API](https://answerline.dev/blog/sync-async-webhooks): When to wait for an answer, when to queue a task and poll, and when to let results come to you by webhook.
- [AI answers now carry ads — here's how to see them](https://answerline.dev/blog/tracking-ads-in-ai-answers): ChatGPT serves ad units inside answers. The ads[] field exposes them — who's advertising, on which prompts, in which markets.
- [How to track what AI assistants say about your brand](https://answerline.dev/blog/tracking-ai-answers): A practical setup for measuring brand mentions and citations in ChatGPT answers, market by market.
- [White-label AI visibility data for agencies and platforms](https://answerline.dev/blog/white-label-ai-visibility): Sell AI-visibility features without building capture infrastructure — one API behind your own dashboards, reports and client portals.
- [Verify webhook signatures before trusting task results](https://answerline.dev/blog/verify-webhook-signatures): Webhook endpoints are public by design — verify the signature, timestamp and event type before storing results. Working verification code in Python and Node.
- [Tracking Google AI Overviews at scale with async batches and webhooks](https://answerline.dev/blog/tracking-google-ai-overviews-async-batches-webhooks): Submit a keyword list as GOOGLE task batches with the AI Overview requested, collect results by webhook, and read the SDKs' typed response.
- [AI-answer alerts without code — a Zapier setup](https://answerline.dev/blog/zapier-ai-answer-alerts): Wire AnswerLine into Zapier so a Slack message or sheet row appears every time a monitored AI answer changes — no backend required.
- [What is generative engine optimization (GEO)?](https://answerline.dev/blog/what-is-geo): A working definition of GEO — the discipline of making your brand appear inside AI-generated answers — and how it differs from SEO.
- [Ask ChatGPT from inside your editor — MCP setup for Cursor and Claude](https://answerline.dev/blog/cursor-mcp-ai-answers): The hosted MCP server turns every engine into a tool your AI editor can call mid-session. One-line setup for Claude Code, Cursor and any MCP client.
- [Track SERP feature changes, not just rankings](https://answerline.dev/blog/serp-features-change-tracking): Featured snippets, local packs, AI Overviews and knowledge panels appear and disappear per keyword. Presence tracking catches what rank tracking misses.
- [Design a webhook receiver that survives outages](https://answerline.dev/blog/webhook-reliability-design): Result delivery over webhooks means your endpoint's downtime is your problem to absorb. Receiver patterns that lose nothing.
- [An AI visibility report clients and executives actually read](https://answerline.dev/blog/ai-visibility-report-template): The structure, metrics and chart set for a monthly AI-visibility report — built from monitor API data, readable by non-SEOs.
- [Catch a brand crisis in AI answers before it reaches the SERP](https://answerline.dev/blog/brand-crisis-ai-monitoring): Negative narratives surface in AI answers — especially Grok's X-fed ones — hours before they rank on Google. A monitoring setup for comms teams.
- [Extending an SEO program to AI answers — the migration map](https://answerline.dev/blog/seo-to-geo-migration): Your keyword lists, rank-tracking cadence and reporting stack all port to answer monitoring. What carries over, what changes shape, what to drop.
- [GEO metrics vs SEO metrics: what maps and what doesn't](https://answerline.dev/blog/geo-vs-seo-metrics): Position, CTR and impressions don't survive the move to AI answers. The metric translation table and what's genuinely new.
- [Design a prompt set worth monitoring](https://answerline.dev/blog/prompt-set-design): The prompts you track define the metric. How to build a prompt set that reflects real buyers — clustering, phrasing, and how many you need.
- [AI answers in every language your customers speak](https://answerline.dev/blog/multilingual-ai-answers): The same prompt in French, German and Japanese returns different answers citing different sources. How to monitor AI visibility beyond English markets.
- [Run your own AI citation study — the honest methodology](https://answerline.dev/blog/run-your-own-ai-citation-study): Most AI-search statistics online are borrowed or stale. How to measure citation rates, trigger rates and source mixes yourself, with a defensible method.
- [Does structured data help AI engines cite you?](https://answerline.dev/blog/structured-data-ai-visibility): Schema.org markup, clean HTML and machine-readable pages — what actually helps an AI engine read and cite your content.
- [AI answers resample — why one check is an anecdote](https://answerline.dev/blog/ai-answer-volatility): The same prompt can name you today and a competitor tomorrow. Answer variance is the core measurement problem in AI visibility — and the fix is statistical.
- [Handle API errors like a pipeline, not a script](https://answerline.dev/blog/api-errors-and-retries): 429s, 402s, timeouts and 5xx each mean something different. The retry policy that respects idempotency — and the one that double-charges you.
- [Estimating your AI-answer monitoring bill before you spend a credit](https://answerline.dev/blog/ai-monitoring-cost-planning): A practical model for sizing prompt sets, engine mixes and cadence so you can predict monthly credit usage before your first run.
- [Sampling vs. census in AI-answer monitoring](https://answerline.dev/blog/sampling-vs-census-ai-monitoring): How to decide how many runs per prompt are enough — the statistics of collecting AI answers without paying for noise.