Fellow

What is Fellow?

Fellow is a meeting management workspace with AI capture. Before a meeting, participants can prepare a collaborative agenda and review previous actions. During or after it, Fellow can record and transcribe the conversation, generate a recap, identify decisions and suggested action items, and connect those records to a searchable library. Ask Fellow uses permitted meeting context to answer questions or help draft follow-up work.

This makes Fellow broader than a standalone recorder. Its value depends on whether the organization uses agendas, notes, assignments, and recurring meeting templates consistently. If every transcript is stored but nobody verifies actions or closes the loop, the library becomes another archive rather than an operating system for meetings.

Recording and transcription routes

The official AI features guide describes support for Google Meet, Zoom, Microsoft Teams, and Slack Huddles, with transcription across many languages. The Note Taker can join based on user or workspace rules, and native capture options are available in supported environments. Uploaded MP3, M4A, WAV, and MP4 files can also produce a transcript and recap.

Fellow's Mac and Windows desktop application offers botless recording. It captures microphone and system audio without adding a participant to the call. This can feel less intrusive, but it does not remove the user's responsibility to tell participants and obtain legally required consent. Official help says the application needs microphone and screen-recording permissions and that internal users see a recording indication in Fellow; external participants may need a separate disclosure.

After processing, the recap can include a summary, transcript, decisions, action items, and speaker labels. Search can find words across transcripts, while Ask Fellow can answer questions over accessible meetings. Check names, amounts, dates, negation, owners, and sensitive statements. A speaker-label error can turn an idea into an apparent commitment by the wrong person.

Agendas, actions, and meeting knowledge

Fellow supports recurring agendas, collaborative notes, templates, feedback, objectives, action items, and integrations with workplace tools. Pre-meeting briefs can surface previous context, and automated recaps can help absent participants catch up. Custom vocabulary can improve specialized transcription under current features.

Use a clear process: the organizer owns the agenda, a named participant approves the recap, and each action has one owner and a date. Avoid automatically emailing an unreviewed recap after a sensitive customer or personnel discussion. Separate brainstorming from approved decisions, and link important conclusions to source moments.

Searchable meeting history can be valuable, but access expands risk. A user asking a cross-meeting assistant a broad question should only receive content they are permitted to see. Test internal, external, private, channel-published, and former-employee scenarios. Define whether managers or workspace administrators can inspect content and how deleted source material affects summaries or AI memories.

Pricing and plan boundaries

Fellow's official pricing guide lists a Free plan, a Solo plan, and Team, Business, and Enterprise options, with a 14-day paid-feature trial that does not require a credit card. Plan names, exact prices, AI credits, and limits can change, so the live page should govern purchasing decisions.

The Free plan is useful for basic notes and a limited number of AI recordings, but lower tiers can limit users, credits, recording access, storage, or integrations. Paid tiers add combinations of more AI notes, automations, project tools, CRM integrations, organization templates, administration, identity, and analytics. Enterprise includes advanced security and retention options under current documentation.

Estimate cost by active users rather than meetings alone. Include administrators, reviewers, integrations, required retention, and users who need Ask Fellow across larger libraries. Measure correction and follow-up time during the trial. Unlimited capture has little value when meetings remain unreviewed or search permissions are poorly governed.

Privacy, training, access, and retention

Fellow states in its Note Taker security and privacy guide that audio is sent to transcription partners, then processed into timestamped speaker-attributed transcripts and summaries. Its official materials say Fellow does not use meeting data to train AI models. This is a meaningful policy statement, but organizations should still inspect the current subprocessor list, contract, data region, retention, and whether partners retain data for abuse monitoring or service operation.

Recordings are limited to meeting attendees by default according to the current Note Taker guide, but users can share recaps by email or publish them to channels. A forwarded link or broad channel can expand access. Review defaults for internal and external meetings and restrict who may publish, download, or change permissions.

Retention is plan-sensitive. Current documentation describes different video-to-audio retention periods by plan and notes that transcripts and summaries can remain with the meeting note. Enterprise administrators can configure automatic deletion for recordings, transcripts, and, when selected, AI summaries. Backup copies may persist for a documented period after deletion. Confirm exact behavior because deleting a video alone may leave a transcript, summary, action, email, integration copy, or exported file.

Alternatives and decision guidance

Fellow fits teams that want structured meeting practice before and after the call. Compare Otter.ai for general live transcription, Fireflies.ai for broad capture and integration coverage, Fathom for a lightweight note-taking workflow, or Read AI for meeting summaries and engagement analysis.

Pilot real internal and external meetings. Score transcript and speaker accuracy after correction, action extraction, agenda adoption, guest consent, permission boundaries, deletion completeness, integration value, and total time saved. The right result is fewer forgotten decisions and lower administrative work without creating an uncontrolled archive of conversations.