Fireflies.ai

Fireflies is centered on collecting a team’s meetings and making their content usable later. Its notetaker says Fred records, transcribes, and summarizes meetings without manual work, while the product pages add searchable transcripts, speaker recognition, action items, and topic analysis. AskFred is the conversational layer over that archive. The service extends beyond the meeting recap with AI Skills that can generate emails, reports, analyses, and scorecards from meeting material. Those outputs are not presented as authoritative decisions. The terms say AI output may be inaccurate and that customers must verify and interpret it before relying on it.

The product also sends material outward after a meeting. Fireflies names CRM, project management, recruiting, and collaboration destinations, including Salesforce, HubSpot, Asana, Trello, Jira, Slack, Greenhouse, and Lever. The integration directory adds storage, dialer, calendar, and document systems. This makes the product more than a recorder, but its primary record remains the meeting transcript, summary, recordings, and related meeting data.

A calendar meeting can be captured by the Fireflies bot, which the vendor says auto joins to record and transcribe. There are also local capture routes. The desktop application supports bot free notes for Zoom, Meet, Teams, and Slack Huddles, and the Chrome extension captures Google Meet calls without a bot. Mobile capture is described for in person conversations, while uploaded MP3, MP4, WAV, and M4A files can also enter the workflow.

This choice matters for teams that do not want a visible bot in every meeting. It does not remove consent responsibilities. The terms put the responsibility for obtaining each meeting participant’s permission on the customer. Fireflies also says participants can see when a meeting is being recorded when the bot has joined.

The entry paid plan is Pro at 18 dollars per user per month when billed monthly. The same Pro card displayed 10 dollars per user per month under annual billing when the page’s billing control was selected. Business is 29 dollars monthly or 19 dollars with annual billing. Enterprise is marked annual only at 39 dollars per user per month. A permanent Free plan exists, but it limits AI summaries and lists 400 minutes of storage per team.

The seat fee is not the entire cost model. Fireflies describes AI credits as a separate pool for advanced AI features. Its billing guide says each AI request uses a dynamic number of credits based on the sources, meetings, or complexity of the request. When included credits are exhausted, additional credits can be purchased, and Auto Upgrade can move an enabled account to another credit tier. The same guide says Auto Upgrade can be disabled in Settings, at which point credit consuming features pause until more credits are available or the next billing cycle begins.

The plan comparison reserves SSO, SCIM, audit logs through the API, custom data retention, and private storage for Enterprise. Private Storage is more specific than a generic cloud claim. Fireflies says Enterprise customers can store transcripts, audio, and summaries in a storage bucket they own and manage, and may choose the storage location during setup. Its private storage guide also states that processing remains on Fireflies servers in the United States.

The company publishes SOC 2 Type II, ISO 42001, GDPR, HIPAA, and FERPA claims. Its public security materials say a healthcare BAA is available on Enterprise with Private Storage. Data handling language is relatively concrete: the terms say User Content is not used to train or fine tune generative AI models, and the privacy policy says account related personal information is deleted within 30 days after account closure. The policy’s zero data retention description refers to third party vendors after meeting content processing, not to immediate deletion from Fireflies itself.

The developer surface is documented rather than merely marketed. Fireflies publishes a GraphQL API that can return selected transcript, speaker, summary, user, and channel data. It also describes webhooks for meeting transcription, summary completion, and notetaker join events. API limits vary by plan. The vendor’s API documentation gives Free users 30 requests a minute and 500 a day, Pro users 60 a minute and 5,000 a day, while Business and Enterprise pool capacity across paid seats.

Its MCP server is aimed at assistants that can query a team’s own meeting content. The API page says Claude, Cursor, and other MCP compatible assistants can search transcripts and pull summaries after connecting to Fireflies. The limits documentation says MCP tool calls share the same limit budget as GraphQL requests from the same user. That makes the MCP interface a read and query route into the meeting archive, with the same plan based capacity constraints as the API.

The terms identify the operator as Fireflies.AI Corp. They grant customers a limited license to use the hosted services and say customers retain rights in their User Content, subject to the service license needed to operate it. The Data Processing Addendum says that, after termination and at the customer’s instruction, Fireflies will delete or return Customer Personal Data unless retention is required by law.

The practical exit route described on product pages is narrower than that contractual instruction. Pro advertises downloads of transcripts, summaries, and recordings. The public materials reviewed here do not describe a full self service export of every workspace object, nor do they describe self hosted deployment or an open source license. Teams with strict migration, hosting, or source access requirements would need a written answer from Fireflies before relying on those capabilities.