Best AI Meeting Minutes Tools in 2026

Meeting minutes are not meeting summaries. They're not transcripts, either. Minutes are structured, formal records of what happened in a meeting: who attended, what was decided, which actions were assigned, and what motions were passed or tabled. They exist to create accountability, satisfy compliance requirements, and give people who weren't in the room an authoritative account of what took place.

Summaries compress a conversation into a narrative overview. Transcripts reproduce it verbatim. Minutes do something different — they extract the governance layer of a meeting: decisions, votes, task assignments with owners and deadlines, and attendee records. If your meetings produce deliverables that get filed, forwarded to leadership, or referenced in audits, you need minutes — not a recap.

The distinction matters because most AI meeting tools were built for summaries first. They got good at producing readable post-meeting overviews with highlights and action items. Whether they produce actual minutes — structured, attributable, exportable records — is a different question. That's what this guide evaluates.

Quick-answer box

Best for structured output closest to formal minutes: Fellow, Fireflies, Avoma

Best free option for lightweight minutes: Fathom

Best for privacy-sensitive environments: Fellow

Best for multilingual teams: Fireflies, tl;dv, Fellow

Important caveat: No tool in this roundup produces true parliamentary-style minutes out of the box. All produce structured summaries that approximate minutes to varying degrees. Teams in legal, board, or compliance-heavy settings should treat AI output as a first draft requiring human review.

How we evaluated these tools

We assessed each tool against criteria specific to the meeting minutes use case, not just general meeting assistant features. Output structure was the most important factor: does the tool separate decisions from discussion? Does it attribute action items to specific people with due dates? Can you identify attendees and their contributions? We also looked at export formats (can you get a Word doc, PDF, or formatted output you can drop into existing templates?), editability of the AI output, template support for different meeting types, and compliance features like recording consent and data retention controls.

Pricing was evaluated based on publicly available information as of early 2026. All monthly prices listed reflect annual billing unless otherwise noted.

The tools

Fellow

Fellow takes a full-lifecycle approach to meetings — agenda, recording, notes, and follow-up all live in one workflow. Its AI generates structured summaries that include action items with assignees and due dates, which maps well to what most teams expect from meeting minutes. The Ask Fellow agent can restructure post-meeting output and answer questions across your meeting history, which is useful for cross-referencing past decisions.

Fellow offers both bot-based and bot-free recording, working across Zoom, Google Meet, and Microsoft Teams. Privacy controls are notably granular: users can pause recording mid-meeting, redact transcript content after the fact, and set sharing permissions by team member — controls that sit with individual users, not just admins. For compliance-sensitive environments, Fellow holds SOC 2 Type II, GDPR, and HIPAA certifications.

The integration ecosystem is deep, with over 50 native connections including Slack, Notion, Linear, Salesforce, HubSpot, Jira, and Asana. This matters for minutes because action items can flow directly into project management tools where they'll actually get tracked.

Where it falls short for minutes specifically: You won't get "motions carried" or "resolutions adopted" formatting without manual editing, but you can create your own AI meeting minutes templates in Fellow to do this.

Pricing: Free plan available (limited AI credits). Team plan starts at $7/user/month. Business at $10/user/month. Enterprise at $25/user/month. AI Note Taker usage may require an additional add-on depending on volume.

Fireflies

Fireflies has built one of the largest meeting intelligence archives in the category, with support for over 100 languages and a searchable transcript library powered by its AskFred AI assistant. Its structured output includes action items, topics, and key decisions, with Smart Search filters that let you quickly locate specific moments across a large volume of past meetings.

For the minutes use case, Fireflies' strength is in post-meeting retrieval and organization. If someone needs to confirm what was decided in a meeting three months ago, AskFred can surface it conversationally. The Topic Tracker feature lets teams flag recurring subjects they want monitored across meetings, which is useful for governance and project oversight.

Where it falls short for minutes specifically: Fireflies' output is organized by topics and speakers but doesn't natively produce a structured minutes document. Action items are detected but attribution can be inconsistent in large group calls. The free plan's 800-minute storage cap and limited AI credits (20 per month) mean serious users will need the Pro plan at minimum. Some users report transcription accuracy can vary with accents and overlapping speakers.

Pricing: Free plan available (800 minutes storage, limited AI credits). Pro at $10/user/month. Business at $19/user/month. Enterprise at $39/user/month.

Otter

Otter pioneered real-time collaborative transcription and remains strong in live note-taking scenarios where multiple participants view and annotate the transcript as the meeting unfolds. Its OtterPilot feature auto-joins calls and generates summaries with action items and key takeaways. The platform supports English, French, and Spanish.

For minutes, Otter's main advantage is the live collaboration layer — team members can highlight, comment, and add context during the meeting, creating a richer document than what AI generates alone. The meeting-level search and organized workspaces help teams maintain a library of meeting records.

Where it falls short for minutes specifically: Otter's output is closer to annotated transcription than structured minutes. Decision tracking and action item attribution are basic compared to tools like Fellow or Avoma. No plan offers unlimited transcription, which is unusual in 2026 — the Pro plan caps at 1,200 minutes monthly and the Business plan at 6,000. Multiple user reviews note inconsistencies in speaker identification and transcription accuracy, and some report a confusing interface. Video replay is restricted to Enterprise.

Pricing: Free plan (300 minutes/month, 30 min per conversation limit). Pro at $8.33/user/month. Business at $20/user/month. Enterprise pricing on request.

Avoma

Avoma is built primarily for revenue teams, but its structured output — which organizes meetings into smart chapters with topic segmentation, action items, and speaker analytics — produces some of the most organized post-meeting documentation in this roundup. The AI notes segment conversations by discussion topic, which naturally maps to a minutes-like structure where each agenda item gets its own section.

The platform also includes built-in agenda management, so you can define meeting structure before the call and have the AI organize its output against that agenda. For recurring meetings with consistent formats, this is the closest any tool gets to producing minutes-grade output without manual restructuring.

Where it falls short for minutes specifically: Avoma's strength is in sales and customer success analytics — talk-to-listen ratios, deal risk scoring, coaching features — which adds cost and complexity for teams that just need good minutes. The conversation intelligence and revenue intelligence modules are separate add-ons. Some users report transcription accuracy varies with audio quality. The sales-centric positioning means general-purpose teams (engineering standups, board meetings) are underserved by the default templates.

Pricing: Free plan with limited features. Starter (AI Meeting Assistant) at approximately $19/user/month. Conversation Intelligence add-on at $29/seat/month. Revenue Intelligence add-on at $39/seat/month. Costs add up quickly if you need the full stack.

MeetGeek

MeetGeek generates structured summaries organized by key takeaways, action items, and highlights, with support for over 50 languages. Its summary templates adapt based on call type — standups, sales calls, interviews — which helps standardize output across different meeting formats. The platform supports automation workflows through Zapier, Make, and n8n, connecting to over 7,000 apps.

For minutes, MeetGeek's template-based approach is useful: you can configure how the AI structures its output and ensure consistency across your organization. The searchable meeting library and custom tagging system help maintain organized records.

Where it falls short for minutes specifically: Users frequently report transcription accuracy issues, particularly with accents, jargon, and industry-specific terminology. The bot joins meetings by default and must be manually disabled when unwanted, which can be disruptive. There's no AI copilot for querying meeting content conversationally. Compared to Fellow or Fireflies, the integration ecosystem is narrower.

Pricing: Free plan available. Pro at $15/user/month. Business at $29/user/month. Enterprise at $59/user/month.

Supernormal

Supernormal positions itself specifically around meeting notes rather than transcription or conversation intelligence, which makes it an interesting fit for the minutes use case. It produces structured AI notes segmented by topics and action items using GPT-4, with a bot-free recording option via Chrome extension. The platform supports over 60 languages and includes a memory feature that connects insights across multiple meetings with the same participants.

Templates are a strength — Supernormal adapts its note format based on meeting type, and the output is designed to be immediately shareable. For teams that need clean, formatted notes after every meeting without extensive post-processing, it's one of the more streamlined options.

Where it falls short for minutes specifically: The output is structured meeting notes, not formal minutes. Customization of export formats is limited. The free plan allows only 20 meetings per month with 1,000 minutes of storage per member. Supernormal requires a Chrome extension for Google Meet, which limits flexibility. Compared to tools with deeper integration ecosystems, the workflow automation options are more basic.

Pricing: Free plan (unlimited meetings, 1,000 minutes storage). Pro at $10–18/user/month (pricing varies by source; check the Supernormal website for current rates). Business at $19–29/user/month.

tl;dv

tl;dv is an AI notetaker with strong support for sales coaching workflows, but its customizable meeting templates and multi-meeting intelligence make it relevant for the minutes use case. You can create templates that structure AI output around specific sections — decisions, next steps, open questions — and the AI will reorganize its notes to match. If you change the template after the meeting, tl;dv will re-structure the notes accordingly.

The platform supports 30+ languages, offers over 7,000 integrations via its automation layer, and includes CRM syncing and sales coaching playbooks (BANT, MEDDIC) for revenue teams. The free plan is notably generous, with unlimited meeting recordings and five file uploads.

Where it falls short for minutes specifically: tl;dv doesn't produce minutes-format output natively. The Business plan ($59/user/month) is expensive relative to alternatives if you don't need the sales coaching and analytics features. There's no mobile app for in-person meetings. The free plan limits AI features, so practical use requires the Pro plan at minimum.

Pricing: Free plan (unlimited recordings, limited AI features). Pro at €18/user/month (~$20). Business at €59/user/month (~$65). Enterprise with custom pricing.

Fathom

Fathom's standout feature is its genuinely unlimited free plan — unlimited recordings, transcripts, and storage across Zoom, Google Meet, and Microsoft Teams, with no monthly caps. AI summaries are limited to five per month on the free tier, but the core recording and transcription functionality is unrestricted, which is rare in 2026.

Fathom generates structured post-meeting recaps with action items and key moments. The output is clean and shareable, and the ability to click on any part of the AI notes and jump to that exact moment in the recording is useful for verifying decisions. HIPAA, GDPR, and SOC 2 Type II compliant.

Where it falls short for minutes specifically: Fathom's output is a structured summary, not minutes. There's no template customization on the free plan, and summaries are formal in tone but not configurable. Collaboration features only appear in the paid Team Edition ($19/user/month) and above. The tool joins meetings as a visible bot participant, which may be inappropriate in sensitive settings. Some users report timestamp inaccuracies and occasional issues with the bot joining meetings uninvited.

Pricing: Free plan (unlimited recordings, 5 AI summaries/month). Premium at $15–19/user/month. Team Edition at $19–29/user/month. Team Edition Pro at $29–39/user/month.

Krisp

Krisp built its reputation on AI noise cancellation and has expanded into meeting notes and transcription. The combination is unique: cleaner audio input means more accurate transcripts, which in turn produces better AI summaries. The tool operates bot-free, running locally on your device, and works across virtually any communication platform — not just the big three videoconferencing tools.

Meeting output includes transcriptions, AI summaries, and action items. Templates let teams structure notes by meeting type. The noise cancellation and accent conversion features make Krisp particularly useful for global teams working in non-ideal audio environments.

Where it falls short for minutes specifically: Meeting notes are basic compared to dedicated meeting intelligence tools. Notes are generated via a manual "Summarise" button with limited customization beyond templates. The AI assistant features are minimal — there's no cross-meeting search or conversational query capability comparable to Fellow's Ask Fellow or Fireflies' AskFred. The free plan limits AI summaries to two per day. Multiple user reviews report reliability issues, including meetings not being recorded or transcriptions disappearing.

Pricing: Free plan (60 min/day noise cancellation, 2 AI summaries/day). Pro at $8/user/month. Business pricing available. Enterprise with custom pricing.

Bluedot

Bluedot is a privacy-focused AI notetaker that operates entirely without bots — it records via a Chrome extension or desktop app, capturing conversations discreetly in the background. It supports over 100 languages and produces AI-generated notes with summaries, action items, and speaker identification. CRM integration with Salesforce and HubSpot is automatic, and there are native connections to Notion and Slack.

For minutes, Bluedot's customizable note templates adapted to specific meeting types (customer calls, all-hands, standups) help standardize output. The annotation and commenting features allow team members to add context to specific moments, enriching the record beyond what the AI generates alone.

Where it falls short for minutes specifically: The free plan is extremely restrictive — five meetings per lifetime. Bluedot was originally Chrome-only and has expanded to desktop apps, but the ecosystem still feels browser-first. User reviews note that customization options for exported summaries could be improved, and some report occasional recording issues and AI inaccuracies. Enterprise-level integrations are limited compared to more established competitors. No Android app as of early 2026.

Pricing: Free plan (5 meetings lifetime, 1-hour max). Basic at $14/user/month. Pro at $20/user/month. Business at $32/user/month.

Comparison at a glance

Fellow — Bot + bot-free recording, 50+ integrations, granular privacy controls, action items with assignees. Free / $7–25/user/month.

Fireflies — 100+ languages, AskFred search, Topic Tracker, large archive support. Free / $10–39/user/month.

Otter — Live collaborative transcription, real-time annotation. Free / $8.33–20/user/month.

Avoma — Agenda-aligned output, smart chapters, strong for revenue teams. Free / $19+/user/month.

MeetGeek — Template-based summaries, 50+ languages, 7,000+ automations. Free / $15–59/user/month.

Supernormal — Bot-free, GPT-4 notes, cross-meeting memory. Free / $10–29/user/month.

tl;dv — Customizable templates, re-structurable notes, generous free plan. Free / €18–59/user/month.

Fathom — Unlimited free recordings, clean summaries, HIPAA compliant. Free / $15–39/user/month.

Krisp — Noise cancellation + notes, bot-free, works across 800+ platforms. Free / $8/user/month+.

Bluedot — Privacy-first bot-free recording, 100+ languages, CRM auto-sync. Free (limited) / $14–32/user/month.

When AI minutes are and aren't sufficient

AI meeting tools have gotten remarkably good at extracting decisions, action items, and key discussion points from conversations. For the vast majority of team meetings — standups, project check-ins, 1:1s, client calls, internal planning sessions — AI-generated structured summaries function as effective meeting minutes with minimal editing.

There are contexts, however, where AI output is a starting point rather than a finished product. Board meetings, shareholder meetings, and any setting governed by parliamentary procedure require a specific format — motions, seconds, vote tallies, resolutions — that no tool in this roundup generates automatically. Legal proceedings, regulatory hearings, and compliance-sensitive discussions need human verification of every recorded decision, because the consequences of an AI misattribution are significant.

In these cases, the most productive workflow is to use AI to generate a comprehensive first draft — capturing all discussion points, decisions, and action items — and then have a human editor restructure it into the required formal format. This still saves substantial time compared to manual minute-taking, but it acknowledges a real limitation: AI can capture what happened, but it can't yet reliably determine what matters in a governance context.

Teams evaluating these tools should also consider consent and recording policies. Some jurisdictions require all-party consent for recording, and the presence of a visible bot (as with Fathom and Fireflies) makes the recording obvious, while bot-free tools (Fellow's botless mode, Krisp, Bluedot) are more discreet but may require separate disclosure. Check your organization's policies and local regulations before deploying any recording tool.

FAQ

What's the difference between meeting minutes and meeting summaries?

Minutes are formal records that document attendees, decisions, action items with owners, motions, and votes. They serve as an official account of what occurred. Summaries are informal narrative overviews that highlight key points and takeaways. Minutes are a deliverable; summaries are a convenience.

Can AI fully replace a human minute-taker?

For most internal team meetings, yes — AI tools produce structured output that captures decisions and action items reliably. For board meetings, legal settings, or compliance-sensitive environments, AI output should be treated as a first draft requiring human review and formatting.

Which tool produces the most minutes-like output?

Fellow and Avoma produce the most structured output, with Fellow's action items including assignees and due dates and Avoma's smart chapters segmenting by agenda topic. Neither produces true parliamentary-format minutes without manual editing.

Do any of these tools work for in-person meetings?

Fellow, Krisp, Bluedot (via mobile app), and Otter support in-person meeting recording through device microphones. Most tools are optimized for virtual meetings on Zoom, Google Meet, or Microsoft Teams.

What about data privacy and where recordings are stored?‍

This varies significantly. Fellow, Fathom, and Fireflies hold SOC 2 Type II certification. Fellow and Bluedot are GDPR compliant. Fellow and Fathom are HIPAA compliant. Krisp processes audio locally on-device for noise cancellation, though meeting notes are generated via cloud APIs. Always review each tool's privacy policy and data retention settings before deployment.

Are free plans actually usable for meeting minutes?‍

If you're looking for a free solution (likely for personal and not work use), Fathom's free plan is the most practical — unlimited recordings and transcripts with no monthly cap. Fellow's free plan includes five AI recordings per user. Most other free plans impose significant limits on transcription minutes, AI features, or storage that make them better suited for testing than daily use.

‍