Fathom AI Notetaker vs Leexi

Fathom AI Notetaker

A bot is documented for Zoom, Google Meet, Microsoft Teams and Slack Huddles with real-time capture, plus a Mac-only bot-free beta that preserves speaker attribution and mute acknowledgment. The captured API page states summaries are 'Always displayed in English', the in-person iOS app is described as coming soon rather than live, and we found no public information on supported transcription languages including German, Webex or phone capture, file upload, custom vocabulary, or transcript editing.

Leexi

Capture covers everything my users touch: Teams, Meet, Zoom and Webex plus eleven telephony sources, an iOS/Android app for in-person meetings needing no room hardware, file upload, live captioning in the dashboard, diarization with speaker naming, and a company dictionary for our jargon, with the speech-to-text engines named (Gladia, ElevenLabs, Azure Speech). We found no public information on accuracy per language, mixed-language meetings, or editing of the transcript, so I stop short of the top band.

Fathom AI Notetaker

Decisions and action items assigned by name are documented and can be pushed to Asana or Notion, with search and Ask Fathom across team calls, trackers, scorecards, deal view and a shared workspace with folders. We found no public information on summary-language selection with German (the API documents English only), templates per meeting type, or answers that cite the transcript passage they rest on.

Leexi

Summaries adapt by meeting type with our own templates and Studio formats, decisions and tasks come out with suggested owners plus a TaskManager, Ask Leexi searches the archive, chapters link back to the recording segment, and minutes export to Word, PDF or Google Docs with an approval workflow for formal meetings. The language models are named (OpenAI via Azure France, Mistral), though we found no public information on answers citing the passage they rest on or on selecting the summary language.

Fathom AI Notetaker

A public REST API with webhooks, TypeScript and Python SDKs, OAuth for installable apps and an official MCP server are documented, alongside HubSpot and Salesforce field sync, Slack, Asana, Notion, Zapier, Make, and Outlook or Google Calendar as a requirement. For a Microsoft shop like mine the gaps matter: we found no public information on posting summaries to Microsoft Teams channels, SharePoint or Confluence export, Microsoft Dynamics or Pipedrive, or recording rules per calendar or meeting type.

Leexi

This is the deep part: Outlook, Google and Exchange on-premises calendars, a dozen CRMs including Microsoft Dynamics 365 and Pipedrive with automated field population, Slack, Notion and SharePoint delivery, Zapier, Make and webhooks, a documented public REST API with reference pages and key management, and a well-specified MCP server with OAuth scopes — more than I need to wire into Microsoft 365. Rules for which meetings the assistant joins are described at the level of connected-calendar settings; we found no public information on rules per calendar or meeting type.

Fathom AI Notetaker

Consent appears as a contractual duty on the user — the terms prohibit use 'if any participant on your call does not consent to be recorded' — and the DPA commits to assisting with data protection impact assessments, with enterprise deletion requests routed through the customer administrator. We found no public information on a recording notice to external guests, a participant opt-out, admin rules for when the assistant joins, whether coaching metrics and AI scorecards can be switched off per organisation, or works-agreement material for a works council.

Leexi

The compliance pages are candid about the law: participants must be informed before recording, the EU AI Act Article 50 duty is explained, default consent is enforced, and admins control when the assistant joins and who may see minutes. For my works council I still miss the rest: we found no public information on a technical opt-out for participants, consent captured from external guests before recording, how to switch off the per-person coaching scorecards and performance analysis, or any works-agreement template or DPIA support.

Fathom AI Notetaker

The enterprise tier documents SSO with SCIM provisioning including Okta, custom data retention policies, admin-configured access and sharing rules, audit logs and regional storage options, backed by SOC 2 Type II, HIPAA and encryption at rest and in transit, and third parties are barred from training on our content with an opt-out even for the vendor's own de-identified training. We found no public information on discarding audio after transcription, named AI and transcription subprocessors with region and zero retention, granular roles, legal hold, or exporting audit logs to a SIEM.

Leexi

The core for my rollout is there: SSO with SCIM, eight roles, centrally set retention (two-year default, reducible per the data processing agreement and a Governance module, with plan-level figures published), advanced audit logs, ISO 27001:2022, encryption in transit and at rest, a public DPA with audit rights and 24-hour breach notice, and contract-end return or deletion. Two things hold me back: the non-training commitment is documented on the privacy policy and compliance pages rather than inside the data processing agreement itself, and we found no public information on discarding audio after transcription, SOC 2, or exporting audit logs to a SIEM.

Fathom AI Notetaker

The contracting entity is Fathom Video Inc., a Delaware corporation in San Francisco under California law, and the privacy policy states the service is 'hosted in the United States'; a DPA on Common Paper standard terms exists with SCCs, Data Privacy Framework certification, an EU representative in Ireland, and a subprocessor list pointed to a trust page that was captured empty. That is a US-hosted processing chain with standard transfer safeguards rather than European storage.

Leexi

Leexi SA is a Belgian company with storage and processing in France on AWS Paris, a Scaleway option above certain licence volumes for strict sovereignty, OpenAI models run via Azure France in a closed circuit, Mistral as a European alternative, and all ten subprocessors listed with their regions, under a published data processing agreement. The Cloud Act caveat on AWS is disclosed openly; transfers outside the EU are said to be possible under standard contractual clauses, and the contact form runs on a US CRM.

Fathom AI Notetaker

Team, Business and Enterprise seat prices are public at $15, $25 and $35 per user per month with a 2-user minimum and Enterprise annual only, plus a free plan with unlimited recordings and no usage caps, published cancellation and auto-renewal terms in US dollars, and non-refundable fees. The annual saving is stated only as 'save 25%+', and we found no public information on what counts as a seat, VAT treatment, storage limits, or the price of add-ons such as Launch Assist — so whether a 400-seat price holds is not derivable from the captured pages.

Leexi

Three of four tiers carry public per-licence prices — "12€/month", "23€/month" and "39€/month", per licence per month, annual invoice, excl. VAT, with volume discounts from 10 licences and minimum users stated — but everything I actually need at 400 seats (SSO, SCIM, audit logs, enhanced retention) sits in the Enterprise tier priced only on contact. We found no public information on what counts as a seat or on monthly billing.