The 11 Best Meeting Notes Tools in 2026 (We Tested Them All)

Key Takeaways

• A transcript is not a record. A transcript proves what was said. A record states what was decided. Those are different documents, and almost no tool ships the second one.
• There are three grades of meeting output: the transcript, the summary, and the decision record. AI notetakers got very good at grade two and stalled at grade three.
• A decision record has four fields: what was decided, who owns it, by when, and what was explicitly rejected. The fourth field is the one every tool here drops.
• Granola ranks first because it treats your typing as the signal and the audio as evidence, the only architecture here that keeps judgment inside the artifact.
• Fathom has 2026's strongest free tier and the most conservative action-item extraction, which is why its lists get worked through instead of ignored.
• Storyflow ranks eleventh and last. No transcription, no recording, no calendar integration, no bot, and it should not be your meeting notes tool.
• Consent rules differ by jurisdiction and by company policy. Ask before you record, and check your own situation rather than a vendor's default.

The Three Grades of Meeting Output

Open the notes from a meeting six weeks ago. Not last week: six weeks. Then answer one question: what did we decide and who owns it now.

Usually you cannot. The document is long, accurate, well organised and useless, because it describes a conversation instead of recording an outcome.

Grade one is the transcript. Every word, timestamped, attributed. Highest fidelity, lowest value, and the fidelity is what makes it unreadable: it preserves the false starts, the tangent about parking, the three minutes where two people talked past each other before agreeing. It is evidence, not information, and its uses are narrow: a disputed conversation, a compliance requirement, a research interview where phrasing is the data.

Grade two is the summary. Two hundred words of prose, a list of topics, and a section called action items. This is where AI notetakers arrived and where nearly all of them stop. Read a good one closely and it says the team discussed pricing and considered several options. It never says which one you picked.

Grade three is the decision record. Four fields per decision, and only four:

1. What was decided, stated as a sentence with a verb.
2. Who owns it, named, one person.
3. By when, a date.
4. What was explicitly rejected, and why.

The fourth field separates a record from a summary. The rejected option is the most valuable line in the document, because six weeks later somebody proposes it again in good faith and the entire meeting happens a second time. A record that says "we considered per-seat pricing and rejected it because our biggest accounts have the most casual users" ends that in ten seconds. A summary that says "the team discussed pricing models" restarts it.

A transcript is not a record. Neither is a summary. The record is a human act, and every tool below is ranked on how close it gets you to performing that act, and on what it does with the action item once it has found one. An action item sitting in a document nobody reopens has not been captured, it has been buried in a nicer filing cabinet.

At a Glance: The 11 Tools Compared

How I Ranked These

I came out of documentary, where a two hour interview yields about ninety seconds of usable film and the discipline is deciding what the material means rather than preserving it. A meeting record needs the same discipline, which is why full-fidelity transcripts look to me like a comfort rather than a solution. Over two years I have run these eleven tools across team meetings, client calls, interviews and workshops.

Five criteria, in order.

1. Does it get past grade two? The test: open a note from a month ago and answer what we decided, who owns it and by when, without opening the transcript.

2. Does it capture the rejected option? The test: name something the meeting considered and did not do, then search the note for it.

3. What happens to the action item? The test: does it reach the system where work is tracked, or stay in a document opened once.

4. Does a bot have to join the call? A visible bot changes the room, particularly with clients and interview subjects.

5. What does it cost per person who needs it? These are priced per seat, and the seat count is always larger than the number of people who will read a note.

Pricing is as of August 2026 and changes frequently. Verify with each vendor.

Consent, Law and Company Policy: Read This First

I am not a lawyer and nothing here is legal advice. Get advice that applies to your situation before you standardise on a recording tool.

What I can say plainly is that recording rules are not uniform. In the United States, consent requirements vary by state: some require only one party to consent and others require every party, and which rule applies can turn on where each participant is physically located rather than where your company is. In the European Union and the UK, a recording and its transcript contain personal data, which brings GDPR obligations around your lawful basis for processing, retention, access, and the rights of the people recorded. Other jurisdictions have their own rules again. Many companies also have internal policies stricter than the law, and some client contracts prohibit recording outright.

The practical position that keeps you out of trouble costs nothing. Ask, out loud, at the start of every recorded meeting, and get an actual answer rather than silence. Say what is being recorded, where it is stored and who can see it. Turn recording off when somebody objects. Do not record 1:1s, performance conversations or anything involving a complaint unless everyone has agreed and your HR or legal function has signed off. Set a retention period and hold to it.

Check your own jurisdiction, your company policy and any client agreement. A vendor's default setting is not a legal opinion, and a tool making something easy is not evidence that you are allowed to do it.

Quick Picks by Job Type

• Best overall: Granola. Your typed notes are the spine and the AI fills in what you missed.
• Best free tier: Fathom. Genuinely usable without paying, which nothing else here matches.
• Best searchable archive: Fireflies.ai. Search across hundreds of past meetings works.
• Best when the note must live next to the work: Notion AI. Items become rows with owners.
• Best live transcript: Otter.ai. Real-time captions and the most mature transcription here.
• Best for research and sales calls: tl;dv. Clip-first, built for sharing thirty seconds.
• Best inside Teams or Workspace: the built-in recap. No bot, no adoption cost.
• Best actual decision record: Obsidian or any plain text file. Grade three, because you wrote it.

1. Granola

The verdict. The only notetaker that treats your typing as the signal and the audio as supporting evidence, which is why its notes read like a person made them.

Best for. Anyone who already takes notes and wants them finished rather than replaced.

Pricing. Free tier covering a limited number of meetings per month. Individual plans run roughly $18 per user per month with higher business tiers, as of August 2026. Mac first, so check platform availability.

Why it ranks here. Every other notetaker assumes the meeting is the input and the human a passive recipient. Granola inverts it. You type during the call, badly, in fragments, and it merges those fragments with what it heard into a document built on your structure.

That sounds cosmetic and is not. Type "pricing: going tiered, Sam pushing back on the enterprise floor" and you have already decided what mattered, so the AI's job shrinks to filling in detail. A tool starting from audio alone over-weights whatever was discussed longest rather than whatever was decided, and the longest discussion is usually the thing nobody could agree on.

Against the four fields, Granola gets the decision and the owner far more often than the pure-AI tools, because you tended to type those, and it comes closest on the rejected option. There is no bot, so nothing appears in the participant list. That is an advantage on client calls and a responsibility: an absent bot does not remove the obligation to say you are recording.

Strengths.

• Your typed fragments shape the note, so the output carries judgment rather than inferred importance.
• No bot in the participant list, so the meeting does not visibly change.
• Notes read like a person's notes, which means people reread them.
• Templates for recurring meeting types genuinely improve consistency.

Limitations.

• Type nothing and the output drops to an ordinary grade two summary, losing the whole advantage.
• Device-level capture records your machine, so hybrid rooms need care.
• Weaker task and CRM integrations than Fathom or Fireflies, and narrower platform coverage.

The trade off. The best notes on this page, conditional on you still doing something during the meeting.

2. Fathom

The verdict. The most trustworthy action-item extraction here, wrapped in the only free tier that is a product rather than a trial.

Best for. Teams whose real problem is follow-ups that get agreed and never happen.

Pricing. Free tier with unlimited recording and AI summaries. Paid individual plans run roughly $19 to $24 per user per month, team plans around $29 to $39, as of August 2026.

Why it ranks here. Fathom is strongest on the back half of the ranking axis: what happens to the action item afterwards. Its extraction is deliberately conservative. A tool listing twelve action items after a forty minute call has not found twelve commitments, it has found twelve future tense verbs, and the list gets ignored because it is padded.

The follow-up mechanics are the other half. It drafts the follow-up email with the items in it and pushes them into CRM records before you have left the call. For someone running six calls a day that is the difference between a CRM reflecting reality and one updated on Friday from memory.

On the four fields it captures the decision and often the owner when ownership was said out loud. It infers deadlines poorly and drops the rejected option entirely, the ceiling every fully automatic tool here shares. The free tier is why a small team can adopt an AI notetaker without a procurement conversation.

Strengths.

• Free tier that a real team can actually run on.
• Conservative extraction produces a list people trust and use.
• Strong CRM and task integrations, so items leave the document.
• Summaries appear within seconds of the call ending.

Limitations.

• A bot joins and appears in the participant list, which some clients dislike.
• No mechanism for your own typed input to shape the note, unlike Granola.
• Weak deadline inference, and it does not capture what was rejected.

The trade off. If follow-through is your failure point rather than note quality, take this over Granola.

3. Fireflies.ai

The verdict. The best archive on this list, which is a different product from the best note.

Best for. Organisations with enough volume that searching across past meetings is a real workflow.

Pricing. Free tier with limited credits. Pro is roughly $10 per user per month billed annually or around $18 monthly, Business roughly $19 annually or $29 monthly, as of August 2026.

Why it ranks here. Fireflies bet that value compounds across meetings rather than within one, and at volume the bet pays. Search your history for a client name, a feature or a competitor and it returns the moments, not just the documents. For a customer success team asking whether this objection has come up before, no amount of single-meeting summarisation replaces that.

The integration surface is the broadest here: items push to task systems and CRMs, and summaries post into channels. It ranks third because the per-meeting note is a competent grade two and no more, reporting what was discussed with good coverage and low judgment.

A searchable archive of every conversation your company has had is an asset and a governance question at once. Who can search it and how long it is kept are decisions to make at rollout, not after the archive is two years deep.

Strengths.

• Cross-meeting search that surfaces the right moment, not just the right file.
• The widest integration set here, including CRMs, task tools and channels.
• Strong speaker separation and speaker analytics on paid tiers.
• Reasonable annual pricing at team scale.

Limitations.

• Bot joins the call and is visible to everyone.
• Action item lists run long, and the single-meeting note is unremarkable next to Granola.
• The archive is a governance obligation, not just a feature.

The trade off. Buy it for the archive and the integrations. Do not buy it expecting a better note.

4. Notion AI Meeting Notes

The verdict. The weakest capture of the AI tools and the strongest destination, because the note lands where the work already lives.

Best for. Teams running projects in Notion who want the decision attached to the project rather than filed beside it.

Pricing. Notion AI is bundled into the Business tier, roughly $20 to $25 per user per month billed annually as of August 2026, with limited AI on lower tiers. Confirm the current packaging.

Why it ranks here. Every other tool here produces a document inside its own application, the structural problem with all of them. The note is correct, complete and somewhere else. Six weeks later you are looking at the project, not the notetaker, and the decision governing it sits in a tool you have not opened since.

Notion's notes land in the workspace where the project database, the spec and the roadmap already are. An action item can be a row with an owner property and a due date property rather than a bullet in prose. That is the most direct route from grade two to grade three any tool here offers, because a row with an owner and a date is structurally a decision record. Notion does not write it for you. It gives the fields somewhere to exist.

The capture is the weakest of the AI options: transcription quality and speaker handling trail the specialists and the summaries are generic. Buying Notion for its meeting notes would be strange while Fathom's free tier exists.

Strengths.

• Notes live beside the project, spec and roadmap they concern.
• Action items become database rows with real owner and date fields.
• No extra tool, subscription or login for teams already on Notion.
• Relations and views can list every decision affecting one project.

Limitations.

• Capture quality trails every dedicated notetaker here.
• Only makes sense if your team genuinely runs work in Notion, and AI sits on the higher tiers.
• Generic summaries that rarely surface the decision without editing.

The trade off. Worse notes, better home. For most teams the home matters more.

5. Otter.ai

The verdict. The most mature transcription product on this list, and the one most firmly stuck at grade one.

Best for. Live captions, accessibility, and situations where the exact wording is the deliverable.

Pricing. Free tier with monthly minute limits. Pro runs roughly $8 to $17 per user per month depending on annual or monthly billing, Business roughly $20 to $30, as of August 2026.

Why it ranks here. Otter has done this longer than anyone and the transcription is excellent: fast, accurate across accents, good at speaker separation, strong on live captions during the call. If somebody in your meeting needs real-time text to participate, Otter is the answer and this post's ranking axis is beside the point.

It ranks fifth because its product philosophy is the one this post argues against. A transcript is not a record, and Otter's centre of gravity is the transcript. Summaries and action items sit on top and behave like it: ask what was decided and you get a paragraph faithfully reporting that the topic was discussed at length. The AI chat moves the problem rather than solving it, since asking about the launch date requires already knowing a launch date was discussed. A record tells you things you did not think to ask.

Otter is also the tool people most often leave running by accident, joining meetings on a calendar rule and building an archive nobody chose. Deploy it deliberately and turn off automatic joining for anything sensitive.

Strengths.

• Best live transcription and real-time captions here.
• Genuine accessibility value during the meeting, not only after.
• Mature, stable and well supported across platforms.
• Good speaker separation and reasonable handling of crosstalk.

Limitations.

• Centre of gravity is the transcript, so summaries feel bolted on.
• Automatic calendar joining creates archives nobody decided to create, and action items are routed nowhere.
• Minute limits on lower tiers bite quickly with daily meetings.

The trade off. Best in class at the grade this post argues is least useful, and still the right answer when accessibility is the requirement.

6. tl;dv

The verdict. Clip-first rather than summary-first, which is the correct shape for research and sales and the wrong shape for a standup.

Best for. User research, customer discovery and sales calls, where thirty seconds of the person beats a paragraph about them.

Pricing. Free tier with recording and transcription. Paid plans run roughly $18 to $29 per user per month as of August 2026, with CRM on higher tiers.

Why it ranks here. tl;dv understood something the others did not: for some meetings the summary is the wrong artifact. In user interviews the point is not what was discussed, it is that a customer said a specific sentence in a specific tone, and no paraphrase carries that weight in a product argument.

Tag moments during the call, pull clips afterwards, assemble a reel of six customers saying the same thing, and drop it into the document where the decision is argued. It ranks sixth because that strength is narrow: nobody clips sprint planning.

Against the four fields, tl;dv does something the others do not, almost by accident. Clipping the moment the team rejected an option is the only practical way here to preserve the fourth field, at the cost of somebody noticing and pressing the button.

Strengths.

• Clip and timestamp during the call, the right primitive for research.
• A thirty second clip beats a summary for changing minds.
• Multi-language support is strong for international research.
• Manual clipping is the closest thing here to capturing a rejected option.

Limitations.

• Overhead for ordinary internal meetings nobody will clip.
• Bot joins the call and is visible, and action item routing trails Fathom on lower tiers.
• Value depends on somebody actively tagging during the call.

The trade off. Buy it for research and sales calls. Do not roll it out company-wide as a general notetaker.

7. Microsoft Teams Intelligent Recap

The verdict. Not the best notes, but zero adoption cost and no bot, which in practice beats a better tool nobody turns on.

Best for. Organisations on Teams, especially where IT will not approve a third-party recorder.

Pricing. Requires Teams Premium at roughly $10 per user per month, or Microsoft 365 Copilot at roughly $30, on top of existing licensing, as of August 2026.

Why it ranks here. Recap is built into the meeting rather than attending it. No bot, no extra participant, no separate tool to learn. It produces speaker-attributed sections, chapters, mentions of your own name, and follow-up tasks inside Teams.

For a large organisation this matters more than note quality. The realistic alternative is not Granola, it is nothing, because procurement will not clear a third-party recorder. A grade two summary every meeting produces automatically is worth more across a thousand employees than a grade 2.5 summary forty of them use.

It ranks seventh because the notes are unremarkable and the licensing is expensive at scale. Chaptering and personal mentions are the good parts: jumping to the ninety seconds where your name came up in a meeting you skipped is a real saving. Governance is why enterprises land here anyway, since recordings stay inside the tenant.

Strengths.

• No bot and no extra tool, so adoption is automatic.
• Recordings stay within your existing Microsoft tenant and policies.
• Chapters and personal mentions make skipped meetings skimmable.
• Follow-up tasks surface inside the Microsoft stack.

Limitations.

• Notes are generic compared with the dedicated tools.
• Requires Teams Premium or Copilot on top of existing licensing, and packaging changes often.
• Only covers meetings held in Teams.

The trade off. The right answer in a Microsoft shop for governance reasons, and the notes are the price.

8. Google Meet Notes

The verdict. The same argument as Teams, one tier down on features and one up on simplicity.

Best for. Workspace organisations that want automatic notes with nothing to install.

Pricing. Included with the Gemini-enabled Workspace business and enterprise tiers, roughly $14 to $22 per user per month as of August 2026. Confirm your tier.

Why it ranks here. Turn it on and afterwards a Google Doc appears in Drive with a summary, discussion points and next steps, shared with participants. It is the lowest friction path to a grade two note that exists.

The Docs destination is underrated. Unlike a notetaker's proprietary web app, the note is a normal document you can edit, comment on, link from a brief and search alongside everything else. Editing matters, because the walk from grade two to grade three is a five minute edit, far likelier in a document you already have open.

It ranks eighth because the summaries are the most generic here and action item handling barely exists: next steps appear as a list and nothing routes them. Attribution trails the specialists on crowded calls. Like the Teams option, its real competition is nothing at all, and an automatic doc beats four people's private half-notes.

Strengths.

• Nothing to install, no bot, no separate account.
• Output is an ordinary Google Doc, editable and linkable.
• Automatically shared with participants and stored in Drive.
• Covered by existing Workspace data governance.

Limitations.

• Most generic summaries on this list.
• Next steps are listed and never routed.
• Availability depends on your Workspace tier, and it only covers meetings held in Google Meet.

The trade off. Free in the sense that you already pay for it, and worth about that.

9. Obsidian and Plain Text

The verdict. The control group, and the uncomfortable one: it captures nothing automatically and produces the highest grade output on this page.

Best for. Anyone willing to spend ten minutes after a meeting for a record still useful in a year.

Pricing. Free for personal use. Sync is roughly $4 to $5 per month and commercial licensing around $50 per user per year, as of August 2026.

Why it ranks here. Ranking a text editor ninth on a list of meeting notes tools needs an explanation, and it is this: Obsidian scores zero on capture and full marks on output. The reason is not the software. It is that a person decided what mattered, and that decision is the entire product.

A four line note reading "Decided: tiered pricing, three tiers. Owner: Priya. Ships: 12 Sept. Rejected: per-seat, because our largest accounts have the most casual users" beats 3,000 words of accurate AI summary, and it takes ninety seconds to type. It hits all four fields including the one no automatic tool reaches. Obsidian adds backlinks, so the project page accumulates every decision that shaped it, and local files, so the record outlives the tool.

The cost is real. It demands discipline immediately after the meeting, when the next one has already started, and most teams cannot sustain it, which is why the AI tools exist. The honest answer is a hybrid: let an AI notetaker hold the transcript, and spend ninety seconds writing the four fields yourself.

Strengths.

• The only approach here that reliably produces all four decision fields.
• Local plain text files outlive any vendor.
• Backlinks accumulate decisions on the project page over time.
• Free, fast, offline, and no consent question because nothing is recorded.

Limitations.

• Captures nothing automatically, so a missed note is gone.
• Requires discipline in the worst five minutes of the day, and nothing is shared unless you share it.
• No transcript to check when memory is disputed.

The trade off. The best output at the highest ongoing cost, paid in attention rather than money.

10. Apple Notes

The verdict. The default most people actually use, now with on-device help, and still fundamentally a private notepad.

Best for. Individual notes on Apple hardware where privacy is the priority.

Pricing. Free with any Apple device. iCloud storage tiers start around $1 per month if you need more space.

Why it ranks here. Apple Notes is here because it is honestly where a great many meeting notes get typed. It has improved: recording and transcription of audio captured on the device, on-device summarisation on supported hardware, and the strongest privacy posture here, since processing need not send your meeting to a vendor.

It ranks tenth because it does nothing about the part of the problem this post is about. No meeting integration, no participant handling, no action item routing, no team-level record. A note living in one person's Notes app is not an organisational record however good it is.

The specific failure to watch for is the one everyone has lived: three people each take notes in their own Notes app, all three sets are reasonable, none is authoritative, and six weeks later there are three different accounts of what was agreed. That is worse than no notes, because it produces confident disagreement.

Strengths.

• Already installed, already used, zero adoption cost.
• On-device processing on supported hardware is the strongest privacy position here.
• Audio recording and transcription of what your device captures.
• Fast, reliable and works offline.

Limitations.

• No meeting or calendar integration of any kind.
• No action item extraction or routing, and private by default, so it produces no team record.
• Apple ecosystem only, which excludes half of most teams.

The trade off. Fine as your personal notepad. It is not a system and should not be mistaken for one.

11. Storyflow

The verdict. Ranks last and deserves to: it does not record, transcribe or attend meetings, and matters only where the output was a diagram rather than prose.

Best for. The workshop or planning session whose real artifact was a board.

Pricing. Paid only during early access. Plus is $7.99 per month billed annually or $9.99 monthly, adding the 200 plus Story blueprints and unlimited file uploads. Pro is $14 annually or $19 monthly, adding AI image generation, roughly twenty times more AI usage and memory across conversations. Max is $39 annually or $49 monthly, adding forty times more AI and Team Workspace with permissions and roles. Pricing is flat per account rather than per seat, and anyone a paid member invites joins free. The Free plan launches before the end of 2026. Pricing as of August 2026.

Why it ranks here. Storyflow is our product and it is last for reasons that are not close. It has no transcription. It has no recording of any kind. It has no calendar integration, no bot that joins a call, no speaker diarisation and no action-item extraction. If you came here for a tool that turns a meeting into notes, ten tools above this one do that and Storyflow does none of it. Go and read the Granola or Fathom review instead.

The narrow ground it has is a different meeting. A workshop clustering forty ideas into six themes, a planning session ending in a sequenced board, an architecture conversation whose outcome is a diagram: for those the transcript is worthless and the AI summary is worse, because the point was spatial and a linear summary flattens it. "The team grouped the ideas and identified several themes" is a true sentence that destroys the artifact.

The second and smaller claim is the board the meeting refers to. When the roadmap or campaign lives on a canvas, the meeting moves the canvas and the canvas is the record. The AI reads your full active canvas board plus up to one Tactic and up to three Documents you @-mention, so you can ask what changed or what is unassigned, which no per-meeting summary answers. A board can hit grade three, because it holds the decision, the owner, the date and the rejected branch beside the chosen one. That is a human act, not something the tool does.

Strengths.

• The right shape for meetings whose real output was a diagram or a clustered board.
• The board the meeting is about can be the record, updated live during the call.
• Whole-board AI answers questions about the plan rather than about one meeting.
• The rejected branch can sit visibly beside the chosen one, which no notetaker here preserves.

Limitations.

• No transcription and no recording. Not a weak version of the feature, an absence of the capability.
• No calendar integration, no notetaker bot and no speaker diarisation, so nothing happens automatically and nothing is attributed to anyone.
• No action item extraction, no task assignment with due dates, no reminders, no follow-up email, and no import from Otter, Granola or Fireflies.
• Realtime multiplayer editing is Team Workspace on Max only, and it is paid only during early access unless a paid member invites you.

The trade off. Use a real notetaker for the meeting. Use a canvas only when what you needed was a picture, and accept that it captures nothing that was said.

What to Actually Pay For

Start on Fathom's free tier and pay nothing. It records, transcribes and summarises without a card. Paying before establishing that anyone reads the notes is the most common waste here.

Pay for Granola when you are already a note-taker. Roughly $18 a month doubles the value of typing you were doing anyway, and is worth nothing if you sit silently and expect the tool to work.

Pay for Fireflies when the archive is the point. The threshold is around fifty meetings a month across the team, or the moment somebody asks whether a client raised an objection before and nobody can answer. Buy seats only for people running meetings that produce commitments.

Do not buy Teams Premium or Copilot for the notes alone. At roughly $10 to $30 per user per month on top of existing licensing, note quality does not justify it. Buy it for governance.

Tools to Avoid for This Job

The always-on calendar bot. A rule joining every meeting builds an archive nobody chose, containing 1:1s, candidate interviews and difficult conversations. It is a consent problem, a retention problem and a morale problem. Turn recording on per meeting.

The shared live doc with no owner. Six people typing into one document produces a beautifully collaborative mess with no single account of the decision. One named notetaker beats six enthusiastic ones.

Chat scrollback as the record. Decisions made in a huddle feel like conversation rather than change, so nothing gets written down. Then the plan is two decisions out of date and whoever missed the thread works from the old one.

Pasting a transcript into a general chatbot and calling it minutes. It works, roughly, and the unredacted contents of a meeting have left your controlled environment for a tool that may not be covered by your data agreements.

Recording a 1:1 because the tool made it easy. Ease of use is not authorisation. Performance conversations, complaints and anything touching somebody's employment should not be casually recorded.

What No Tool on This List Does

None of them capture what was rejected. Every tool here can tell you the team discussed three options. Not one records which two were killed and why, the field that stops the same meeting happening again.

None of them make the decision. A meeting that ends without one still produces a beautiful summary of an unresolved conversation, and the tooling makes that ambiguity look tidy.

None of them own the follow-up. Fathom and Fireflies push an item into a task system, the strongest thing on offer, and an item with no owner and no date still will not happen.

And Storyflow, our own product, does not capture a single word of the meeting. No transcription, no recording, no calendar hook, no bot, no attribution, no action items. It is eleventh because for the job in the title it does almost nothing.

The Bottom Line

The AI notetaker category solved the wrong half of the problem, brilliantly. Transcription is close to solved and summarisation is very good, and neither has changed the fact that teams cannot answer what they decided six weeks ago.

Use Granola if you type during meetings, Fathom if you do not and want the follow-up chased, and Fireflies if the archive is where your value sits. If you live in Teams or Workspace, turn on what you already pay for. Ask for consent every time and set a retention period before the archive gets deep.

Then do the ninety seconds nobody wants to do. Write the four fields: decided, owner, date, rejected. Put them where the project lives, not where the notes live. A transcript is not a record, and no tool here, including ours, will write the record for you.