canonical: https://jentic.com/apis/fireflies.ai/fireflies

# Fireflies.ai GraphQL API

Jentic publishes the only available OpenAPI specification for Fireflies.ai GraphQL API, keeping it validated and agent-ready. Fireflies.ai is an AI meeting assistant that records, transcribes, and summarises meetings across Zoom, Google Meet, Microsoft Teams, and Webex. The API is GraphQL-based with a single POST /graphql endpoint that exposes meeting transcripts, speaker turns, AI-generated summaries, action items, sentiment, AskFred conversation threads, channels, users, analytics, and audio/video upload mutations.

## For AI agents

Query meeting transcripts, summaries, action items, and AskFred conversation threads, and upload new audio/video for transcription - all through a single GraphQL endpoint.

## Scope

Does not handle calendar scheduling, video conferencing, or live-meeting bot control - use for transcript, summary, action-item, and AskFred operations only.

## Capabilities

- Query meeting transcripts with speaker turns, timestamps, and confidence scores
- Retrieve AI-generated summaries, action items, and key topics for a transcript
- Search transcripts across an organisation by keyword, attendee, or date range
- Start and continue AskFred conversation threads for question-answering over meetings
- Upload audio/video files for asynchronous transcription via the GraphQL upload mutation
- Manage users, channels, and team analytics through dedicated GraphQL types

## Use cases

### Meeting Action-Item Extraction

Pull action items, owners, and due dates from finished meetings using the action_items GraphQL field. Pipe them into a task tracker so nothing falls through the cracks. Because everything goes through POST /graphql with one bearer token, integration is straightforward - most teams have an action-item bot live in a day.

Example prompt: Query the GraphQL endpoint for transcript(id: 'm_abc') { action_items { text assignee dueDate } } and create matching tasks in Asana

### Conversational Q&A Over Meetings

Use AskFred via the GraphQL API to ask natural-language questions about a meeting or set of meetings. The conversation thread persists, so an agent can follow up with 'and what did they say next?' style questions. Fireflies handles retrieval and answer synthesis on the meeting transcript.

Example prompt: POST /graphql with mutation askFred(meetingId: 'm_xyz', question: 'What objections did the customer raise?') and return the answer to a Slack thread

### Cross-Meeting Search and Analytics

Search transcripts across an organisation for keywords, attendees, or topics, and pull team-level analytics like talk-time and sentiment trends. Useful for sales coaching dashboards, customer-insight digests, and recurring research summaries. Live in 1-2 days.

Example prompt: Query transcripts(filter: { keyword: 'pricing', dateAfter: '2026-06-01' }) { id title attendees }, then summarise the most-mentioned objections

### AI Agent Meeting Operations

Use Fireflies through Jentic so an agent can answer 'what did we agree in the standup?' or 'send me the summary of yesterday's customer call'. Jentic exposes the GraphQL operations as discoverable tools, with the bearer token in the vault. Get started with Jentic One, the self-hosted execution layer

Example prompt: Through Jentic, search 'get meeting transcript and summary', load the GraphQL operation schema, and execute with meetingId='m_2026_06_09_standup'

## Key endpoints

| Method | Path | Description |
| --- | --- | --- |
| POST | /graphql | Single GraphQL endpoint for queries and mutations across transcripts, summaries, action items, AskFred, users, channels, analytics, and uploads |

## Key resources

- **Transcript** — Meeting transcripts with speaker turns, timestamps, and confidence
- **Summary** — AI-generated summary, key topics, and sentiment per meeting
- **ActionItem** — Extracted tasks with assignee and due date
- **AskFred** — Conversation thread for question-answering over meetings
- **Upload** — Audio/video upload mutation for asynchronous transcription
- **User** — Users in the organisation and their meeting attendance
- **Channel** — Logical groupings for meeting series
- **Analytics** — Talk-time, sentiment, and engagement analytics

## Why Jentic

- **Setup:** Wiring the Fireflies.ai GraphQL API by hand means setting its bearer auth, composing GraphQL queries and mutations against a single endpoint, and handling retries yourself. Through Jentic you install once, import Fireflies from the API Directory, store the token once, and your agent calls it.
- **Permission scoping:** Fireflies exposes a single GraphQL endpoint and selects its target through query variables rather than an owned resource in the URL path, so limit the agent to the operations it needs, such as fetching a transcript or its action items, and leave the rest out of the allowed set. You choose which operations it may call.
- **Credential handling:** Your Fireflies token is stored once, encrypted, by your own Jentic One instance and injected into the Authorization header at execution time. It never enters the agent's prompt, logs, or context.
- **Discovery method:** Agents search Jentic by intent such as 'get meeting action items' or 'ask a question about a meeting', and Jentic returns the matching Fireflies GraphQL operation with the variables it needs so the agent calls the right query without browsing the reference docs.

## Related APIs

- **Fitbit Web API** — Provides personal health context to pair with Fireflies meeting summaries
- **Firebase Auth REST API** — Adds user authentication to a Fireflies-powered meeting analytics app
- **Firmwater LMS API** — Turns sales-call insights into training assets in an LMS

## FAQ

### Why is there no official OpenAPI spec for Fireflies.ai GraphQL API?

Fireflies.ai exposes a GraphQL API and does not publish an OpenAPI specification. Jentic generates and maintains this spec so that AI agents and developers can call Fireflies.ai GraphQL API via structured tooling. It is validated against the live API and kept up to date. Get started with Jentic One, the self-hosted execution layer.

### What authentication does the Fireflies.ai API use?

The API uses HTTP bearer authentication with a Fireflies API key (generated in your Fireflies account settings). Through Jentic, the bearer token is stored in the encrypted vault and added to the Authorization header at execution time.

### Can I get meeting action items with the Fireflies.ai API?

Yes. Query the transcript GraphQL type and request the `action_items` field; each item includes the text, assignee, and due date when one is detected. The single endpoint is POST /graphql.

### What are the rate limits for the Fireflies.ai API?

Fireflies applies per-account rate limits on the GraphQL endpoint and quota on transcription minutes per plan. The endpoint returns standard 429 responses when limits are exceeded.

### How do I summarise a meeting through Jentic?

Search Jentic for 'get meeting summary', load the GraphQL operation schema, and execute with the meetingId. The response includes the summary, action items, and key topics in a single round trip.

### Does Fireflies support uploading recordings for transcription?

Yes. The GraphQL API exposes an upload mutation for audio and video files, returning a meeting record once transcription completes asynchronously. Webhooks notify your system when a transcript is ready.

### Can I limit what my agent is allowed to do with the Fireflies.ai GraphQL API?

Yes. Because you run Jentic One yourself, your own rules decide which Fireflies operations and credentials the agent can use. Since Fireflies exposes a single POST /graphql endpoint that picks its target through query variables rather than a resource in the URL path, you scope access at the operation level, allowing only the calls the agent needs, such as fetching a transcript or its action items, while leaving mutations like uploading recordings out of the allowed set. The bearer token stays with your instance and the agent can only invoke the operations you approve.