{{locationDetails}}
{{locationDetails}}
Facts and pricing verified .
Fathom tops this ranking for a boring reason, and boring is exactly what you want here. Its free plan carries unlimited recordings and transcriptions, and the official site states SOC 2 Type II, GDPR (the EU privacy regulation) and HIPAA (the American health-privacy rule) compliance. The rating is 5.0/5 across 6,668 G2 reviews.
But Fathom sends a bot into your call, and it records virtual meetings only.
That bot is the whole fight in this category, and Read AI shows you why. Same product, two audiences: 4.0/5 from 43 paying G2 reviewers, and 1.3/5 from 107 reviewers on Trustpilot. What sits in that gap is the bot. Trustpilot's reviewers describe one that joins without clear consent and then resists removal, and those complaints dominate the low scores, alongside privacy worries and cost at scale.
Is a third-party participant appearing on a customer's Zoom a problem for your legal team? Then start with Granola, Bluedot or Jamie. And if your data has to stay in the EU, start with Jamie or MeetGeek.
Answer that one, and the EU one with it, and your list is already shorter. Here's the awkward part of shopping what's left. Most serious products below claim SOC 2 Type II and CRM (customer relationship management) sync, many now ship an MCP server, and nine of the twenty-two advertise a language list in the hundreds. MCP is the Model Context Protocol. It lets Claude or ChatGPT query your meeting history directly.
So the feature grid has stopped separating anyone. What still separates these tools is how each one enters your room, and how its meter runs once it gets there.
Those six are where to start, not where to sign. Accuracy drops on accents, crosstalk and background noise for nine of the twenty-two tools here, on the evidence their own reviewers leave. That is the most common complaint in this field, with slow or unresponsive support running close behind it. So which of them holds up on your audio? No product page can tell you. Answer that one yourself. Two weeks is enough.
Rule of thumb: if a step in that list is awkward at five seats, it gets worse at 300.
That pilot runs on somebody's free plan, and free is where your rollout gets expensive. Free in this category spans two orders of magnitude. Unlimited recordings at one end. Five notes that never come back at the other.
Some plans renew every month. Some hand you a one-time allowance for the life of the account, which is a very different thing to give a team of 200. Read the middle column first.
| Tool | What the $0 plan gives you | Renews or lifetime | First thing that forces the upgrade |
|---|---|---|---|
| Fathom | Unlimited recordings and transcriptions, instant AI summaries, clips, playlists, search | Ongoing | Advanced summaries and AI action items sit on Premium at $20/month |
| Granola | Basic plan at $0/user/month | Ongoing | Limited meeting history; unlimited history needs Business at $14/user/month |
| Fireflies.ai | Core transcription and summary features | Ongoing | The free-tier caps are unclear; Pro at $10/month per user billed annually brings 8,000 storage minutes and two-hour meetings |
| MeetGeek | 3 hours of transcription per month, 2-hour meeting limit | Monthly | The 3-hour ceiling; Pro at $9.99/user/month lifts it to 20 hours |
| Fellow | 5 AI notes and 5 AI recordings | Lifetime of the account | Your sixth meeting; Team at $7/user/month annually gives 10 notes and 10 recordings per user per month |
| tl;dv | Free plan at EUR 0.00/month | - | Recording time limits on longer sessions; Pro is EUR 29.00/month per seat, or EUR 18.00/month per seat billed annually |
| Bluedot | 5 meetings | Lifetime of the account | Your sixth meeting; Basic at $14/user/month is audio-only with a 1-hour cap per recording |
| Jamie | 10 meetings per month, 30-minute maximum duration | Monthly | The 30-minute cap; Plus at EUR 21/month billed annually raises it to 2 hours |
| Notta | 120 transcription minutes per month, 3 minutes maximum per recording, 1 seat | Monthly | The 3-minute recording cap, which no real meeting fits inside |
| Grain | 20 meetings per month; free viewer seats can view and collaborate but not record | Monthly | The 20-meeting cap, the limitation reviewers cite most |
| Tactiq | 5 AI credits per month, 10 transcripts per month, 1 user | Monthly | The 10-transcript cap; Pro at $8/user/month billed annually unlocks unlimited transcripts |
| Plaud | 300 transcription minutes per month | Monthly | The 300-minute ceiling; Pro at $17.99/month gives 1,200 minutes |
| Laxis | 300 minutes per month, 30-day conversation history | Monthly | The 30-day history window closing on your older calls |
| Circleback | Unlimited meetings, AI notes, action items, plus API, MCP and command-line access | Ongoing | 30-day history and a 2-team limit; Pro at $15/user/month billed annually lifts both |
| Supernormal | 15 credits per month, 5 per day maximum, no credit card | Monthly | The 5-credit daily ceiling on a busy day |
| Read AI | 5 meeting transcripts per month, 1-hour maximum, no audio or video playback | Monthly | Playback, which starts at Enterprise ($22.50/month annually) |
| Otter.ai | 300 transcription minutes per user per month, 20 AI Chat queries per month, 3 lifetime file imports | Monthly, except the 3 imports | The 20-query AI Chat cap; Pro at $16.99/user/month raises it to 50 |
A dash means the vendor does not publish it.
Which of these could a team of 200 actually live on? Two. Fathom and Circleback hand you an ongoing plan with no monthly ceiling on the thing you came for, and history you can keep. Fellow and Bluedot hand you five, and five is all you ever get. Both ends of that get the same word on the pricing page.
Avoma, Sembly AI, Spinach.io, Krisp and Gong sit outside the table. Avoma gives you a 14-day trial of the Organization plan with every add-on enabled and no credit card. Krisp gives you seven days. Sembly retired its old free Personal plan, and Gong has no free tier at all. Spinach.io shows a free-start call to action and no published plan names, prices or limits, so treat it as a quote conversation.
A free tier only matters if the paid tier above it clears your security review. SOC 2 Type II is table stakes now. Most of this list claims it, so it has stopped being a differentiator and become a checkbox.
Your real gate sits one layer down. Which plan carries the control your reviewer is asking about? Budget that plan, whatever the marketing page leads with.
| Tool | Named certifications | HIPAA starts on | Identity controls start on | Data residency | Training on your data |
|---|---|---|---|---|---|
| Fathom | SOC 2 Type II, GDPR, HIPAA | Stated, no tier named | Enterprise (custom): SSO and SCIM | - | - |
| Granola | SOC 2 Type II, GDPR with data processing agreement | - | Enterprise, $35/user/month: SSO | - | Third-party AI providers may not train on customer data |
| Fireflies.ai | SOC 2 Type II, GDPR, HIPAA | Stated, no tier named | - | - | Never used for AI training; 0-day retention policy |
| MeetGeek | SOC 2 Type II, HIPAA business associate agreement, GDPR, CCPA | Stated, no tier named | Enterprise (custom): SSO and SCIM | EU and United States | Never used to train foundation models |
| Fellow | SOC 2 Type II, HIPAA, GDPR, CCPA | Stated, no tier named | Enterprise, $25/user/month billed annually, minimum 10 users | - | Never used to train AI models |
| tl;dv | SOC 2 Type II, GDPR | - | - | - | No customer data used to train the AI; metadata shared with Anthropic is anonymized |
| Avoma | - | Enterprise, $39/seat/month billed annually, minimum 10 seats | Enterprise, $39/seat/month billed annually, minimum 10 seats | - | - |
| Bluedot | SOC 2 Type II audit, GDPR | - | Business, $32/user/month: SSO and SCIM | - | Never used to train or improve AI models |
| Jamie | ISO 27001, GDPR | - | Enterprise (custom, 10+ users): SAML SSO and SCIM | EU only, hosted in Germany | No customer data used for model training; audio deleted after transcription |
| Notta | SOC 2, ISO 27001, GDPR, HIPAA, APPI, CCPA | Stated, no tier named | - | - | Enterprise offers an AI-learning-disabled option |
| Grain | SOC 2 Type II, GDPR | - | - | - | Not shared with or used to train third-party AI models |
| Sembly AI | SOC 2 Type II, Microsoft 365 certified, GDPR | MAX, $39/user/month | - | United States and EU | Enterprise Plan customer data not used for model training |
| Tactiq | SOC 2 Type II, ISO 27001, GDPR, HIPAA | Stated, no tier named | Business, $29.17/user/month billed annually: SAML SSO | - | Audio not retained; OpenAI API data excluded from model training |
| Spinach.io | SOC2 Type II, GDPR, HIPAA | Stated, no tier named | - | - | Data encrypted and excluded from AI model training |
| Plaud | SOC2, HIPAA, GDPR | Stated, no tier named | - | - | Not used to train models unless you explicitly opt in |
| Laxis | - | - | Enterprise (custom): SSO | - | - |
| Krisp | SOC 2, GDPR, HIPAA, PCI-DSS (the payment-card security standard) | Enterprise (custom) | Enterprise (custom): SSO and SCIM | - | Transcripts and recordings stored in Krisp Cloud by default; on-device handling is an Enterprise add-on |
| Circleback | SOC 2 Type II, EU-U.S. Data Privacy Framework, HIPAA with signable business associate agreements, GDPR-aligned data processing agreement | Enterprise, $35/user/month billed annually | Enterprise, $35/user/month billed annually: SSO and SCIM | - | Does not use customer data to train models |
| Supernormal | - | - | Business, $40/month: SSO | - | - |
| Gong | - | - | Okta-based provisioning and SSO | - | - |
| Read AI | SOC 2 Type 2, GDPR | Enterprise+, $29.75/month annually, 5 licenses minimum | Enterprise+, $29.75/month annually, 5 licenses minimum: SAML and SCIM | - | - |
| Otter.ai | - | Enterprise, as an add-on | Enterprise (custom): SSO and SCIM | - | - |
Again, a dash means no published value on record.
Now add up the column nobody adds up. The tier carrying identity controls lands between $25 and $39 per user per month across this field, mostly on annual billing, from Fellow Enterprise at $25 billed annually with a ten-user minimum up to Avoma Enterprise at $39 per seat billed annually with a ten-seat minimum. Check the billing term on each before you compare them side by side. Supernormal sits outside that shape: its Business plan is priced at $40/month rather than per seat.
That's two to four times the entry price the same vendors advertise. And fourteen more won't put a number on it at all. Fathom, MeetGeek, Jamie, Laxis, Krisp, Gong and Otter route identity controls to a quote, and Fireflies.ai, tl;dv, Notta, Grain, Sembly AI, Spinach.io and Plaud publish nothing at all. Budget a conversation either way.
Which answers the question you probably walked in with. Can the free plan your team is already sitting on become the company standard? A free plan can run a pilot. It cannot survive an infosec review that asks for SSO. Fathom is exactly that case: unlimited recordings at $0, with SSO and SCIM waiting on a custom Enterprise quote.
That table settles what you pay. It says nothing about who notices the tool in the room. So ask the question your legal team will ask you. Does a stranger appear in the participant list of a customer call?
That one answer filters this market faster than any feature comparison, and it doesn't track the ranking at all.
Tip: bot-free isn't automatically the safer choice for you legally. It just moves the disclosure job from the meeting invite onto your team.
That disclosure job is the last gate. Here is the field measured against all of them. One note on the ratings below: a score means nothing without the count behind it, so every rating comes with its review volume and its platform. A 4.8/5 from 21 reviews and a 5.0/5 from 6,668 are not the same claim wearing different numbers.
Fathom does the whole category job on your Zoom, Google Meet and Microsoft Teams calls. Transcripts, AI summaries, action items, clips, playlists, and Ask Fathom to search across your past calls.
What puts it first is the gap between what it charges and what it hands you. The $0 tier includes unlimited recordings and transcriptions. No expiry, no countdown.
Does an unlimited free tier survive an infosec review? No. Fathom's SSO is a quote.
Past free, the ladder climbs fast. Premium is $20/month ($16/month annually). Team is $19/month per user ($15/month annually, minimum 2 users). Business is $34/month per user ($25/month annually). Enterprise is custom, and that's where SSO, SCIM provisioning and custom data retention live.
If you run sales, the deeper end sits on Business: CRM field sync, deal view, coaching metrics and AI Scorecards.
Salesforce, HubSpot, Slack, Notion, Asana, Zapier, Make, ChatGPT, Claude and a public API. That connector set is one of the widest here. Fathom states 300,000+ companies use it, and G2 named it #1 AI Notetaker with #1 product satisfaction in 2024.
Now the caveats, and they're specific. Users report bugs sharing recordings and granting your team access, including one slow support response on an account email change. Several say the bot is awkward to add to ad-hoc, last-minute meetings. Accuracy drops when your speakers talk over each other.
And Fathom is virtual-meeting only. Everything that happens in a room is out of scope.
Granola's gate is meeting history. Basic at $0/user/month keeps it limited, Business at $14/user/month makes it unlimited and adds advanced integrations and API access, and Enterprise at $35/user/month is where SSO and admin controls start. Regular use pushes you onto Business fast.
The product listens to computer audio and never joins your call. Your rough typed notes come back as structured summaries, action items and follow-up drafts. Calendar-aware Briefs land before the meeting, AI chat across your meetings comes after it, and there is mobile and Apple Watch capture too.
Granola raised a $125M Series C led by Index Ventures (Danny Rimer) in March 2026 at a $1.5B valuation, up from $250M at its May 2025 Series B, as TechCrunch reported. Granola names PostHog, Intercom, Figma, Brex, Linear and Asana among its users.
Vendor viability is one box this fills that few tools its size can.
Granola publishes a trust posture to back the money up: SOC 2 Type II, GDPR with an available DPA (data processing agreement), encryption at rest and in transit, and a policy that third-party AI providers including OpenAI and Anthropic may not train on customer data.
The evidence is thin, though. G2 shows 4.8/5 from 21 reviews, 90% of them five-star. Inside that small base, the flags are speaker-identification gaps in multi-speaker calls and the absence of a fully-featured web app.
Fireflies claims 95% transcription accuracy. The named AI layer on top is unusually specific. Fred is the meeting bot. AskFred answers questions across your meeting content. Live Assist coaches you in real time, with sentiment and topic tracking running underneath.
Its integrations page lists 100+ connected apps. The reach goes past the usual CRM and video suspects into recruiting, with Greenhouse, Lever and BambooHR, plus dialers and storage.
Fireflies documents one of the fuller compliance stories here: SOC 2 Type II, GDPR, HIPAA, AES-256 at rest, TLS in transit, a 0-day retention policy, and a stated commitment that your data is never used for AI training. Fireflies states 1M+ companies use it.
Fireflies publishes its prices in its knowledge base. Free is $0. Pro is $10/month per user billed annually ($18 monthly), giving you up to 8,000 storage minutes per seat and meetings up to 2 hours. Business is $19/month per user annually ($29 monthly) and adds API access, analytics, private channels and video recording. Enterprise is $39/month per user annually.
The review base is deep on both platforms: 746 reviews on G2, rating it 4.7 to 4.8/5, and the same count again on Trustpilot. The recurring complaints across both are accuracy falling with heavy accents or overlapping speech, plus integration friction.
Carry one flag into your read of those reviews. Trustpilot has noted a review-solicitation pattern that can bias the visible sample.
MeetGeek publishes both EU and United States data residency, and its security page is detailed. SOC 2 Type II, a HIPAA business associate agreement, GDPR, CCPA, AES encryption at rest through AWS KMS, development aligned to OWASP ASVS 4.0.2, and an explicit commitment never to use your data to train foundation models.
The product leans automation rather than archive. Meeting Skills are event-driven rules that fire after a call to update your CRM, create your tasks and draft your emails. You also get 30+ summary templates, automatic meeting-type detection, Voice Agents for lead qualification, and analytics covering 100+ KPIs.
Native connectors cover HubSpot, Salesforce, Pipedrive, Notion, Asana, ClickUp, Slack and Google Drive, and Zapier, Make and n8n open 10,000+ more. MeetGeek states 50,000+ teams across 100+ countries.
Reviewers describe transcription and summary quality that "saves me literally days of work," and one puts the saving at 30 to 60 minutes of writing up minutes per call.
The review base is small, and that's the honest brake on all of it. Capterra shows 4.5/5 from 11 reviews, Trustpilot 4.6/5 from 28, so 39 in total. Inside that base the repeated complaints are accuracy on difficult audio, slow or unresolved support tickets with no live chat, and AI Chat search limited to recent meetings on lower tiers.
Pricing runs Basic free with 3 hours of transcription per month, Pro at $9.99/user/month with 20 hours and $0.5 per extra hour, Business at $17/user/month with unlimited transcription and a 3-hour meeting limit, and custom Enterprise carrying unlimited transcription, custom video retention and SSO with SCIM.
Fellow takes the enterprise slot because it ships the controls a security review actually asks about.
Fellow names SOC 2 Type II, HIPAA, GDPR and CCPA certification. An explicit statement that your meeting data is never used to train AI models. Then the unusual part: botless recording, zero-day retention, MNPI redaction and password-protected recaps.
For a finance, legal or healthcare team, that last group is the difference between a pilot and an approval.
Salesforce, HubSpot and Pipedrive for CRM. Asana, Jira, Linear, ClickUp, Monday, Trello and GitHub for work. Okta, OneLogin, Workday and BambooHR for identity and HR. That reads like your IT stack rather than a sales toolkit.
Capterra reviewers rate it 4.9/5 across 37 reviews, describing setup in about five minutes and support replies inside five minutes. Against that, they report loading delays, occasionally missing notes, and deleted action items you cannot recover.
tl;dv earns its place on evidence. G2's 530 reviewers put it at 4.7/5, 84% of them five-star, praising time-coded summaries and multilingual transcription.
It records your Zoom, Google Meet and Microsoft Teams calls automatically. You get customizable AI summary formats built for sales workflows, including Smart AI Topics, plus trend analysis and call coaching across your meeting library.
The security commitment reads solid: tl;dv documents SOC 2 Type II, GDPR, AES-256 at rest, SSL and HTTPS in transit, no customer data used to train the AI, and anonymized metadata where Anthropic is involved. The company states over 2 million users.
Then the plumbing. That same G2 base reports broken or limited HubSpot and Slack integrations, paired with slow support on exactly those tickets. It's a rough combination when the integration is the reason you bought.
That pairing is what holds tl;dv at six.
Pricing is in euros: Free at EUR 0.00/month, Pro at EUR 29.00/month per seat (EUR 18.00/month per seat billed annually, EUR 216.00/year), and Business at EUR 39.00/month per seat (EUR 29.00/month per seat billed annually, EUR 348.00/year). Recording time limits come up as a constraint on longer sessions, and accuracy varies in crosstalk-heavy discussion.
Avoma is your pick if you want one vendor covering more than notes. It runs four product areas: the AI Meeting Assistant, a Scheduler with Lead Router, Conversation Intelligence and Revenue Intelligence.
The integration set is deep where it counts, with six native CRMs (Salesforce, HubSpot, Pipedrive, Copper, ZenDesk Sell, Zoho), eight video-conferencing tools, multiple dialers, Slack, ClickUp and an API platform.
Sentiment is high and well-evidenced. G2's 1,362 reviewers put it at 4.6/5 with 75% five-star, a Call Recording sub-score of 9.5 and Quality of Support at 9.3. The same base reports accuracy varying on difficult audio and occasional notetaker drop-offs. Avoma reports serving 1,000+ organizations.
Then your bill. Base plans are Startup at $19/month annually ($29 monthly) per recorder seat, up to 25 paid seats. Organization is $24/month annually ($39 monthly) up to 100 seats, adding custom AI notes, group scheduling and API access.
What does the platform cost once the add-ons are on? Add $19 to $29 a month, three times over. Conversation Intelligence, Revenue Intelligence and Lead Router are all separate line items on top of the seat.
So the breadth that makes Avoma attractive is what makes your quote climb. There's no permanent free tier here either, only a 14-day trial of the Organization plan with all add-ons enabled and no credit card.
Now the gate. HIPAA, SSO and DPAs all sit together on Enterprise, at $39 per seat billed annually only, with a 10-seat minimum. That's the number to take into your review, not the $19 on the front of the page.
Bluedot puts its gate on the Business tier, at $32/user/month. CRM and ATS (applicant tracking system) integrations, SSO, SCIM and custom retention all live up there, so your compliance upgrade and your integration upgrade arrive on the same invoice.
Below it, Free is 5 meetings for the lifetime of the account. Basic at $14/user/month is unlimited but audio-only, with a 1-hour cap per recording. Pro at $20/user/month adds video and drops the duration cap. Enterprise is a custom, fixed company-wide price regardless of user count.
It takes that pick for capturing without a bot in the call, with numbers that hold up across two independent listings. G2 shows 4.8/5 from 226 reviews. The Chrome Web Store shows 4.7/5 from 200 ratings behind 80,000+ users. On G2, the praise keeps naming the same two things: transcription accuracy, and the fact that nothing visibly joins your call.
Capture is bot-free across Zoom, Google Meet, Microsoft Teams, Webex, Slack and in-person, with speaker identification and technical-terminology support.
The security page states a SOC 2 Type II audit, GDPR-compliant controls, AES-256 and TLS encryption, backups cleared within 30 days of deletion, and no customer data used to train or improve AI models. Bluedot states it is trusted by 50,000+ companies and is backed by Google.
Two complaints repeat. The product feels primarily optimized for Google Meet, and export summary customization is limited.
Model the $32 tier from day one. For anyone syncing a CRM, that's simply what Bluedot costs.
Has your data protection officer drawn a line at the EU border? Jamie is the shortest route across it.
Capture is bot-free across Zoom, Teams and Meet, plus your in-person conversations, and speaker memory carries recognition between your calls. Integrations sync to HubSpot, Salesforce, Attio, DealCloud and Affinity, convert items into Asana tasks, and push notes to Notion, Google Docs or OneNote.
Trustpilot puts Jamie at 4.7/5 from 42 reviews, 91% five-star, with praise for the unobtrusiveness and the support. The same base reports occasional recording glitches and missed participants once a meeting passes 20 attendees.
Now the ladder, because Jamie steps by duration as much as by volume. Free at EUR 0/month gives you 10 meetings with a 30-minute maximum. Plus at EUR 21/month annually gives 20 meetings at up to 2 hours. Pro at EUR 39/month annually gives unlimited meetings at up to 5 hours. Team is EUR 33/month per seat for 2 to 10 users, and Enterprise is custom for 10+ users.
So price Jamie against your longest call. That clock decides your tier before your headcount does.
How long can a single free Notta recording run? Three minutes.
Above that gate, Pro is $13.61/month ($8.17/month billed annually at $97.99) for 1,800 minutes and 5 hours per recording. Business is $27.78/month ($16.67/month annually at $199.99) for unlimited. Enterprise starts from 51 seats.
The product itself covers capture from more angles than most. Meeting bots for your online calls. Bot-free desktop recording. A hardware device called Notta Memo. File and YouTube import.
Claimed accuracy is 98%+, across 58 languages for transcription and translation. Set those two numbers beside each other before you commit, because 58 falls well short of the hundreds that several vendors here advertise. Check yours is on it.
Notta Brain answers questions across your meeting history, there is a bilingual translation layer, and the platform auto-generates presentations, infographics and documents. Compliance is broad on paper: Notta lists SOC 2, ISO 27001, GDPR, HIPAA, APPI and CCPA, with an AI-learning-disabled option on Enterprise. Notta states 10M+ users.
G2's 234 reviewers put it at 4.4/5, praising speed and an interface your non-technical staff can use, while reporting accuracy loss in noisy rooms, with overlapping speech, or with strong accents. They also want deeper API documentation.
Back to that 3-minute cap. No real meeting fits inside it, so the free plan is a look at the interface and nothing more.
Grain's own G2 reviewers recommend it as the cheaper answer to Gong-class tooling, though Gong publishes no price to check that against. It holds 4.6/5 from 299 reviews with 80% five-star, and Starter costs $19/seat/month.
Capture is bot-optional, using computer audio or a joining bot, with speaker recognition. Ask anything searches your meeting library and returns citations. One click exports enriched Markdown transcripts into Claude or ChatGPT. Native CRM sync covers Salesforce, HubSpot and Pipedrive, and you get Slack, Zapier and a public API.
Grain's trust page is more concrete than most at this size. It documents:
Grain claims 97% of teams remain active after 12 months. Its Capterra listing carries no reviews at all, though, so every rating above rests on one platform. Business-tier pricing is missing from the pricing page too, so a sales conversation comes before a number.
Then read the Enterprise condition before you plan a phased rollout. Every seat has to sit on the Enterprise tier. There's no putting a few power users up there and leaving everyone else where they are.
Sembly's angle is what you get out the other end. It generates your proposals, status reports and project briefs from meeting content, rather than a transcript with bullets underneath. Multi-meeting AI Chat queries your whole workspace history, so old decisions stay findable.
G2 rates it 4.6/5 across 46 reviews, with an AI Text Summarization sub-score of 9.2. Sembly states it has processed 2 million+ business interactions for 10,000+ professional teams, and carries Gartner "Emerging Visionary" recognition. It lists SOC 2 Type II, Microsoft 365 certification, GDPR, and United States and EU data residency.
The free Personal plan is gone, which changes your evaluation path. You get a capped trial, then a bill.
Basic is $17/month ($10/month annually) for one user, with unlimited meetings, up to 5 hours of media upload and one year of history. Pro is $29/user/month ($20/month annually) for up to 40 users, with 5 AI-generated documents and 5 AI insights. MAX is $39/user/month ($30/month annually) for up to 500 users, with 40 documents, 40 insights and unlimited history.
G2's reviewers flag cost, speaker-attribution errors, and friction in search and editing.
And note where HIPAA sits: on MAX, the top named tier, at $39/user/month. If HIPAA is your gate, $39 is Sembly's real entry price, whatever the pricing page opens with.
Cheap, light, no bot in the call. So what's the catch? Billing.
Start with what works. Tactiq runs as a Chrome extension across your Google Meet, Zoom and Microsoft Teams calls, with real-time transcription and speaker identification, and it connects to fifteen named tools:
Tactiq states that audio is never retained, only text, alongside SOC 2 Type II attestation, ISO 27001, GDPR and HIPAA, and says OpenAI API data is excluded from model training. It also states 1 million+ active Chrome extension users and 1 million+ meetings transcribed weekly.
The pricing ladder is granular. Free is $0 with 5 AI credits and 10 transcripts a month for one user. Pro is $8/user/month annually with unlimited transcripts. Team is $16.67/user/month annually for 1 to 20 users. Business is $29.17/user/month annually for 20 to 200 users, and that's where SAML SSO starts. Enterprise is custom above 200 users, with a dedicated account manager and a curated DPA.
Users also report no mobile support, and manual language adjustment needed per meeting for non-English audio.
Now the catch in full. On Trustpilot the complaint is billing after cancellation, one reviewer describing charges continuing for over a year, alongside an unresponsive chatbot and email support.
That base is only 15 reviews. Weigh it accordingly, then cancel a seat yourself during the pilot and watch what your card does.
Spinach.io fits you if your meetings produce tickets. It transcribes and summarizes, then creates your tickets, sends recap emails and updates your CRM records.
Capture runs across Zoom, Google Meet, Microsoft Teams and Slack. Downstream you get Trello, Linear, Notion, ClickUp, Asana, Confluence, Monday, Jira, HubSpot, Zoho, Salesforce, Zapier and Attio.
G2's 25 reviewers rate it 4.6/5 and describe it doing the documentation job of a scrum master. Setup is the thing they single out.
Easy quick setup and very logical. No support needed.
The company states that your data is encrypted and excluded from AI model training, and names SOC2 Type II, GDPR and HIPAA. Spinach.io is operated by StayIn, Inc. and lists Y Combinator, Zoom and Atlassian as backers, an unusual investor line-up for a tool this size.
The complaints are specific. Summarization accuracy declining over time, particularly with names. A subscription price called steep for startups and bootstrapped teams. Missing pieces the same reviewers want, including ChatGPT-style question-answering over their notes and whiteboard capture.
That price complaint arrives without a number attached to it. Spinach.io shows a free-start call to action and no published plan names, prices or limits, so treat the whole evaluation as a quote conversation. With 25 reviews behind the rating, treat the sentiment as directional too.
Plaud is a different shape of product. It sells wearable and handheld recording hardware paired with an AI transcription and summarization platform.
So it captures what a notetaker bot sitting in a video call cannot reach: the corridor conversation, the client lunch.
The software layer gives you 10,000+ summary templates, AutoFlow automation, a Plaud Agent, Skills and Connectors, and Ask Plaud, which returns answers you can trace back to their reference in the recording. Exports run to 27+ formats. Plaud says Plaud Intelligence runs on GPT-5.5, Gemini 3.1 Pro and Claude Sonnet 4.6 across all subscription tiers.
Sentiment is strong and well-evidenced. Trustpilot shows 4.7/5 from 2,479 reviews, 85% of them five-star, on top of a claimed 4.9/5 App Store rating from 20,000+ ratings and a stated 2.5 million+ professionals using the products. The praise lands on accuracy and speaker separation.
The complaints land somewhere software buyers rarely have to think about. Slow support against a marketed 24-hour response. Hardware defects such as charging and display failures, paired with frustrating warranty and replacement processes.
That's the tradeoff when you buy a device.
Compliance is stated as SOC2, HIPAA and GDPR, with your conversation data used for model training only if you explicitly opt in. The named integration set is narrow: Google Calendar, plus a Device SDK and API.
Laxis captures your Zoom, Google Meet, Microsoft Teams and Webex calls bot-free, and the connector list is long for what it costs.
On top of transcription you get AI Chat across your meeting history, 50+ report templates, real-time in-meeting guidance and cross-meeting trend analysis, plus a separate voice-keyboard dictation product and OSO earbuds and 360 camera hardware. Founded in 2020, Laxis states 100,000+ professionals across 4,000+ organizations.
Pricing is friendly. Basic is $0/month with 300 minutes and 30-day conversation history. Premium is $9.99/month billed yearly with 2,000 minutes and unlimited storage. Business is $19.99/month billed yearly with unlimited transcription, CRM automation and that 7,000+ app reach. Enterprise is custom, with full API, SSO and private AI models. There is a 14-day Premium trial and no credit card.
The 4.8/5 G2 rating rests on 30 reviews, and inside them the complaints are unresponsive support and development on foundational features described as stagnant. So that number holds the product up less firmly than it looks.
So why does a connector set this wide land at sixteen? The trust page. If your infosec review opens by asking for a SOC 2 report, check Laxis's row in the gate table before you shortlist it.
Krisp comes at this from the audio side. Its noise cancellation is the established product, and the AI Meeting Assistant rides the same subscription. You get bot-free recording with synced playback, action-item assignment, speaker identification, and transcription and summaries in 17 languages at a stated 96% accuracy.
Seventeen is the shortest list among the vendors here that publish a language count. Check yours is on it before anything else.
Your integrations cover Zoom, Google Meet, Microsoft Teams, Slack Huddles, Salesforce, HubSpot, Pipedrive, Affinity, Asana, Jira, monday.com and Zapier. Krisp states SOC 2, GDPR, HIPAA and PCI-DSS, and G2 shows 4.7 stars across 1,543 reviews at the company level.
Is call clarity already a problem for your team? Then the bundle is the argument. If it isn't, read the caveats first. There is no permanent free tier, only a 7-day trial.
Core is $8/month per user annually ($16 monthly) with 1 hour a day of accent conversion and 10 GB of storage. Advanced is $15/month per user annually ($30 monthly) with 4 hours a day of speaker accent conversion, unlimited listener-side conversion, and 60 GB. Enterprise is custom, carrying unlimited storage, SSO, SCIM, HIPAA and private on-device transcription.
On older or heavily loaded machines, G2 reviewers report CPU lag, choppy audio and fans spinning up during real-time suppression. They also report accuracy drift on non-English audio and speaker attribution slipping in busier calls.
Rule of thumb: a privacy control sold as an add-on is a control you don't have yet. By default your transcripts and recordings live in Krisp Cloud. Device-only handling arrives with Enterprise, and nowhere below it.
Circleback's free tier is startling. Zero dollars per user per month buys unlimited meetings, AI notes and action items, plus API, MCP and command-line access, which several vendors here reserve for a paid tier.
Ask Circleback searches across everything you have recorded, with speaker identification on the transcripts, and there are 20+ native integrations covering Slack, Notion, Linear, HubSpot, Salesforce and Asana.
A free tier that good usually buys a higher rank than eighteen. Not here, and the distance between the two lines below is why.
Those reviewers report the app crashing or failing to capture meetings, in one case for extended periods. Support is described as a scripted chatbot with no route to a human or a ticket. One called the setup the worst user experience they had seen.
A strong feature list with almost no independent corroboration is a specific kind of risk, and that's what this ranking is penalizing.
Paid tiers are Pro at $15/user/month and Business at $25/user/month, both billed annually. Enterprise is $35/user/month billed annually, and that one line carries unlimited recording storage, SSO, SCIM and HIPAA together. If any of those four is a requirement, the free tier stops being the story.
Supernormal captures bot-free from your desktop app, then turns your meeting context into more than notes. AI Slides, AI Documents, AI Spreadsheets and AI Images, with customizable templates.
That output range suits agencies producing client deliverables from calls. Supernormal states 700k+ organizations, and G2 rates it 4.4/5, on a review base of about 40, with praise for auto-assigned action items and a responsive product team.
Credits, not seats. Supernormal switched models in April 2026, so this is newer than most of what you will read about it.
Credits are harder to forecast than a flat minutes limit, so map one real month of your usage onto them before you sign.
Reviewers also note there's no single consolidated action-item view across meetings.
Gong is a revenue intelligence platform that captures meetings on the way past, and it sells to a different buyer than most of this list. It states 5,000+ customers including Fortune 10 companies, naming Anthropic, ADP, HubSpot, LinkedIn and Uber for Business. That roster is the buyer definition.
Its product page covers automatic capture of your interactions, the Revenue Graph, the Revenue Harness for AI agents, an AI Trainer for coaching scenarios, forecasting and multilingual support. Native integrations reach your Zoom, Microsoft Teams, Salesforce (bi-directional sync), Microsoft 365 Copilot, Slack and Okta. Gong states it ranked #1 in 17 G2 categories and topped G2's Revenue Operations and Intelligence category.
Ratings depend where you look. G2's listing shows 4.7 stars across 6,671 reviews. Gong's own homepage cites 4.5/5 across 6,200+. Reviewers value the AI call summaries and the recording depth, report inconsistent transcription accuracy on technical terminology, and more recently raise rising price and perceived data lock-in.
Is meeting capture the reason you would buy Gong? Then you're buying a revenue platform to get it.
Your real obstacle is forecasting the cost. There is no published price and no free tier. Licenses are priced per user with a platform fee, quoted against four team-size brackets: 1 to 50, 51 to 1,000, 1,001 to 9,999, and 10,000+. Read that ladder as a picture of who Gong is built for. Standardizing notes across 20 to 500 people puts you at the bottom of a purchase designed for the top.
Read AI is genuinely feature-rich. Ask Read searches across your meetings, email and chat and returns citations. You also get an Ada Digital Twin personal assistant, sentiment and engagement tracking, speaker coaching and video highlights. There are 30+ native integrations including Salesforce, HubSpot, Asana, Notion, Atlassian, Google Calendar and Zapier with webhooks.
The company states SOC 2 Type 2 certification and cites coverage in Fortune, TechCrunch, Forbes and Reuters. Paying G2 customers rate it 4.0/5 across 43 reviews.
Trustpilot's 107 reviewers put it at 1.3/5.
That gap opened this article, and here is what sits inside it. Reviewers report the meeting bot joining without clear consent and persisting afterward, resisting removal even after repeated account deletion attempts. One reported it summarizing private conversations with lawyers and tax and financial advisors.
Read that sentence again with your recording-consent policy in hand.
Still shortlisting it? Then know the tier map.
Otter is probably why you're reading this. It spreads bottom-up, a rep installs it, and then someone asks whether it can be the company standard.
Can it clear an infosec review? Not as it stands.
Otter's security page speaks in general terms about security, governance and compliance controls rather than naming certifications, and HIPAA with SSO and SCIM is an Enterprise add-on. If you're standardizing 20 to 500 seats, that's your whole conversation, and it has nothing to do with the transcription.
Because the transcription is fine. OtterPilot auto-records, transcribes and summarizes with decisions and action items. Otter AI Chat answers questions across your meetings, Channels organize by team or topic, and live transcription covers multiple languages with speaker recognition.
Zoom, Google Meet, Microsoft Teams, Slack, Salesforce, HubSpot, Google Docs, Dropbox, Notion, Jira, Asana and Glean all connect. Your capture can run without a bot, through the desktop app, Chrome extension or mobile. G2 rates it 4.4/5 across 462 reviews.
G2's reviewers report inconsistent speaker identification in group settings, accuracy dropping with noise, crosstalk or strong accents, and costs escalating once usage passes the free tier.
Three run on with no monthly ceiling on the core job. Two of them are generous enough to run a team on.
Fathom, Granola and Circleback all carry ongoing free plans with no monthly cap on the core job. Fireflies.ai's free plan is ongoing too, but its caps are unclear, so treat it as an unknown rather than an unlimited. The two you could actually build on are Fathom's, with unlimited recordings and transcriptions, and Circleback's, with unlimited meetings, AI notes and action items inside a 30-day history window.
Fellow and Bluedot are the opposite animal. Five notes and five meetings respectively, for the lifetime of the account rather than per month. The ten in between renew every month with a cap attached, which is a third promise again.
But the question underneath is whether an ongoing free plan can become your company standard, and the honest answer is no, not the moment identity controls are a requirement. SSO sits on a paid or quoted tier at every vendor here. So pilot free, then budget the tier that clears your review.
Yes. Eight capture without a bot by default, and Notta's desktop recording makes nine.
Look at what you're trading, though. Removing the bot removes the visible cue your participants were reading, so your disclosure practice has to cover what the software no longer announces. Write that wording before the rollout. Someone will ask why they were recorded, and the answer should already exist.
Grain, tl;dv and Otter.ai let you choose per call, which is the useful middle when some of your meetings are internal and some are not.
Price the tier, not the badge. A compliance line in a footer tells you nothing about your invoice.
Four vendors put a price on the plan HIPAA starts on, and those cluster tightly: Read AI's Enterprise+ at $29.75/month annually with a five-license minimum, Circleback's Enterprise at $35/user/month billed annually, Sembly's MAX at $39/user/month, and Avoma's Enterprise at $39/seat/month billed annually with a 10-seat minimum. Call it $29.75 to $39 per user per month, and budget accordingly.
Krisp puts HIPAA on a custom Enterprise plan, and Otter sells it as an Enterprise add-on. Both are quote conversations.
Fathom, Fireflies.ai, Fellow, Notta, Tactiq, Spinach.io and Plaud state HIPAA compliance at the company level, and MeetGeek offers a HIPAA business associate agreement. Ask each of them which tier carries it before you commit a number to your budget.
Only two turn it into a setting you control. Notta offers an AI-learning-disabled option on Enterprise, and Plaud uses your conversation data for training only if you explicitly opt in.
Everywhere else it's a policy line. Fireflies.ai, MeetGeek, Fellow, tl;dv, Bluedot, Jamie, Grain, Spinach.io and Circleback all state that customer data is not used for AI training, and Fireflies adds a 0-day retention policy. Granola states that third-party AI providers may not train on customer data. Tactiq stores text only and states OpenAI API data is excluded from training. Sembly states Enterprise Plan customer data is not used for model training.
A policy is something you enforce in a contract. A switch is something you enforce yourself. If that difference matters to your reviewer, put it in the DPA.
Bluedot, Jamie and Granola capture your in-person conversations alongside virtual calls. Plaud takes the hardware route with wearable and handheld recorders. Notta ships the Notta Memo device plus bot-free desktop recording.
Fathom is virtual-meeting only, and that's the single biggest limitation on an otherwise leading product.
Three, and each one costs you something different.
Jamie hosts and processes in Germany only, holds ISO 27001 and deletes audio after transcription. Its free tier caps meetings at 30 minutes, and SSO with SCIM sits on Enterprise from 10 users up.
MeetGeek publishes EU and United States residency on its security page, with no tier named against it, alongside SOC 2 Type II, a HIPAA business associate agreement, GDPR and CCPA. Its Enterprise tier is where unlimited transcription, custom video retention and SSO with SCIM sit.
Sembly AI supports United States and EU residency too. There is no free tier to pilot on, so your evaluation starts as a bill.
Notice what all six of those questions did. Every answer you give removes names from the list. Here is the order to ask them in.
Start with the room. Decide whether a third party may appear in a customer's call, because that one answer rules out the five tools that send a bot into the meeting, and puts a condition on the three that offer one, before you compare anything else.
Whatever survives that goes to procurement next. Find the plan carrying your compliance control and price that plan, because the tier with identity controls on it runs $25 to $39 per user per month across this field, largely on annual billing. The front of the pricing page leads with $0 to about $20.
Then run what is left against a real week of your calls, and watch where it stops.
Fathom is the default for most teams, and its free tier means your pilot costs nothing but calendar time. Two things take it off your list: a third party in a customer's call, and an SSO line that arrives as a quote rather than a price.
Fellow is the one to walk into a security review with. For consistency, Bluedot. Want the platform rather than the notetaker? That's Avoma, if your budget can absorb the add-ons. And when the data cannot leave the EU, Jamie decides it for you.
And if a notetaker already arrived at your company without anyone approving it, your honest first move isn't choosing a replacement. It's finding out what is already recording.