Fathom AI Notetaker vs Leexi

Fathom AI Notetaker

The bot covers Zoom, Microsoft Teams and Google Meet plus Slack Huddles, a botless Mac beta is documented alongside it with speaker attribution and mute acknowledgment kept, and capture, transcribe and summarize run in real time. But I found no public information stating German among the supported languages, and the API reference says the summary is always in English — a real question mark for my German-language calls — while Webex, phone calls, file upload, timestamps, custom vocabulary and transcript editing are not described on the pages I read.

Leexi

The assistant joins Teams, Zoom, Google Meet and Webex, VoIP services like Aircall and Ringover bring in phone calls, and the mobile app plus upload covers in-person and file capture — everything my eight reps need. Speaker diarization, a company dictionary for jargon and over 100 automatically detected languages are documented, and the speech-to-text engines are named (Gladia, ElevenLabs, Azure Speech), though we found no public information on per-language accuracy, mixed-language meetings or an editable transcript.

Fathom AI Notetaker

This is close to my wish list: a three-sentence summary with the decisions made and each action item assigned by name, a ready-to-send follow-up email after every meeting, and tasks pushed to Asana or Notion, plus search, Ask Fathom across team calls, trackers, Deal View and a shared workspace with folders. It stays below the top band because summaries are stated to always be in English, and custom templates per team, answers that link back to the transcript passage, and the models behind the summaries are not described.

Leexi

Summaries adapt per meeting type with our own templates and formats, decisions and tasks come out with suggested owners, Ask Leexi searches the archive for commitments, minutes export to Word, PDF and Google Docs, audio excerpts can be shared, and the summary models are named (Azure OpenAI, Mistral). We found no public information on a follow-up email draft or on answers that cite the transcript passage they rest on, and both matter for my follow-up discipline.

Fathom AI Notetaker

The HubSpot piece is exactly what I need: CRM field sync maps AI-extracted pain points, timelines and next steps into Salesforce and HubSpot records after every call with no manual entry, on top of a documented REST API with webhooks and TypeScript and Python SDKs, an official MCP server, Zapier, Make, Slack, Asana and Notion, with Outlook and Google Calendar required. What holds it back: no Pipedrive or Microsoft Dynamics, no Confluence or SharePoint export, and no field mapping per meeting type or recording rules per calendar are described.

Leexi

This is the part I care about most and it is documented: notes and tasks sync to HubSpot, Salesforce, Pipedrive and Dynamics with automated CRM field population, Slack, Notion and SharePoint receive minutes, Zapier, Make and webhooks connect the rest, and a documented REST API plus an MCP server are public. The gap is that per-meeting-type CRM mapping and recording rules per calendar or meeting type are only partially described.

Fathom AI Notetaker

The terms put the consent duty on every user, the bot joins as a visible participant, the botless beta keeps mute acknowledgment, and the DPA promises assistance with data protection impact assessments. But I found no recording notice to external guests, no participant opt-out mechanism, coaching metrics and AI scorecards with no stated way to switch them off per organisation, and nothing at all on works-council material, a works-agreement template or the EU AI Act — a gap I would have to close myself before any German rollout.

Leexi

The vendor states participants must be informed before recording begins, cites the EU AI Act transparency duty, lets admins control when the assistant joins, and recordings are deletable on request — but we found no public information on a notice to external guests, an opt-out mechanism, works-council material or DPIA support. The coaching module sells evaluation scorecards and performance analysis, and we found no public information on per-person analytics being off by default, which is exactly what I do not want hung over my team.

Fathom AI Notetaker

Admin-level retention controls, regional storage options and audit logs, SOC 2 Type II with HIPAA and penetration testing, SSO with Okta SCIM provisioning, encryption at rest and in transit, and contract terms barring third parties from training their models on our calls with an opt-out for the in-house de-identified training — a governance story I can take to our data protection officer. Discarding audio after transcription, granular roles, and the subprocessors with their regions and retention are not documented in the pages I read.

Leexi

Retention defaults to a maximum of two years and can be reduced by the client, a governance module brings SSO with SCIM, eight roles, retention policy and advanced audit logs, encryption in transit and at rest is stated with ISO 27001:2022, and the commitment not to train on conversations is stated in the privacy policy and compliance FAQ rather than as a clause of the data processing agreement. We found no public information on discarding audio after transcription, SOC 2 or legal hold, and support confirms there is no one-button export of the whole history.

Fathom AI Notetaker

The contracting entity is Fathom Video Inc., a Delaware corporation in San Francisco, the service is hosted in the United States, and the safeguards are Standard Contractual Clauses plus the Data Privacy Framework, with an EU representative in Cork and the Netherlands named for EEA transfers — a US vendor with a published DPA, not a European one. The subprocessor list is only pointed to at the trust center without names on the captured pages, and which language model and transcription providers read our recordings, and where, is not stated.

Leexi

Leexi SA is a Belgian company, recordings, transcripts and minutes sit in France on AWS Paris with a Scaleway option above certain licence volumes, and ten subprocessors are listed with locations, including Mistral and Azure OpenAI in the EU, with the OpenAI model run via Azure France in a closed circuit. A public data processing agreement, a written Cloud Act notification policy and ISO 27001 are documented; we found no public information on SOC 2, and the vendor itself notes AWS is theoretically subject to the Cloud Act while offering Scaleway for strict sovereignty.

Fathom AI Notetaker

The tiers are public — Team "$15 /month / per user (2 user min)", Business "$25 /month / per user (2 user min)", Enterprise "$35 /month / per user *Annual only", and a free plan with unlimited recordings and no usage caps, with "Annually (save 25%+)" and a 90 day guarantee. But what counts as a seat, the VAT treatment for a German buyer, storage limits and which plan the "Max 3 users/ domain" CRM limit applies to are unclear, and Launch Assist is flagged as a paid add-on without a price.

Leexi

Three tiers are public at 12, 23 and 39 euros per licence per month with annual invoicing excluding VAT, plus a seven-day free trial, minimum user counts, volume discounts from ten licences and unlimited meetings and summary storage on the paid tiers. The Enterprise tier is priced on request, and we found no public information on monthly billing, on what counts as a seat beyond "per licence", or on the price of add-ons such as SSO, the governance module or French cloud hosting.