Jamie vs Otter.ai

Jamie

Capture is botless on the user's own device across Zoom, Teams, Google Meet and any other platform, with in-person recording from laptop and phone, speaker identification, optional speaker memory that names speakers in later meetings, and custom vocabulary by plan; the pages also state that audio is streamed for transcription and never stored. What holds it back is that accuracy per language is undocumented, German is never named among the "99+ languages", and we found no public information on live transcription, timestamps, phone or dialer calls, or file upload.

Otter.ai

A bot joins Zoom, Teams and Google Meet while a desktop app and a Chrome extension record without a bot, in-person capture runs on iOS and Android, calls arrive via Zoom Phone, RingCentral and Salesloft, and German sits among six languages with custom vocabulary, editable time-coded transcripts and live captions. We found no public information on Webex, mixed-language meetings or accuracy per language — and the terms expressly make no warranty about the completeness or accuracy of the transcription, which is exactly the clause a regulated team should read before trusting a word.

Jamie

Summaries come structured with an executive and full-by-topic breakdown, editable templates that auto-apply to the meeting type, tasks with assignees that can be pushed to Asana, search across meetings, and an "Ask Jamie" chat over the archive. My standing question is unanswered: the model that writes the summary is never named, nothing documents that a generated statement links back to the transcript passage it rests on, and choosing the summary's output language is not described.

Otter.ai

Templates per plan, action items pushed into JIRA, Asana and ClickUp with deadlines, follow-up emails drafted in Work Mode, keyword and sentiment tracking across meetings, and an AI chat over the whole archive, with Anthropic and OpenAI named on the subprocessor list for AI functionality. But I always ask which passage a summary sentence came from: we found no public information on answers that cite and link the transcript passage they rest on, on the summary language being selectable, or on how hallucinations are handled.

Jamie

A documented REST API with reference material, stated rate limits, two key types, webhooks and a request builder, an MCP server, post-meeting workflows that update the CRM, and native integrations spanning HubSpot, Salesforce, Dynamics 365, Attio, Affinity, DealCloud, Notion, OneNote, Google Docs and Asana. We found no public information on Slack or Teams channel delivery, Zapier, Make or n8n, or SharePoint and Confluence export, and authentication is a static API key rather than scoped OAuth — with an API base URL that still says beta.

Otter.ai

Calendars auto-join meetings, Salesforce, HubSpot, Microsoft Dynamics, Zoho and Monday.com take the CRM output, Slack, Notion, SharePoint and OneDrive receive the notes, Zapier and viaSocket automate further, an API with webhooks ships on Enterprise, and an MCP server for AI assistants is listed from the Basic plan up. We found no public information on API rate limits, versioning, OAuth scope granularity or whether residency survives the export, so treat this as a rich catalogue rather than a documented platform contract.

Jamie

Two notice mechanisms are documented — a pre-meeting email to attendees with timing windows, domain exclusions and Enterprise enforcement, plus a chat notice posted once per recording — and the pages state plainly that informing participants remains the user's responsibility. The chat notice is beta and off by default, we found no public information on a way for participants to stop or opt out of a recording themselves, works-council material, DPIA support and the EU AI Act classification are not documented, and the optional speaker memory processes biometric voiceprints that a works council would want in writing.

Otter.ai

The terms put the whole burden on the buyer: the customer is "solely responsible for providing any notices to, and obtaining consent from, individuals in connection with any recordings". Sentiment and engagement monitoring across customer-facing teams is sold as a feature with no public statement that it is off by default or switchable per organisation, and we found no public information on a participant opt-out, a recording notice to external guests, works-council material or DPIA support — nobody external to the room is protected by the product itself as documented.

Jamie

Audio is streamed and deleted immediately after transcription, no model training on customer data is stated on every plan, ISO 27001 is described as independently audited and certified, SAML and SCIM arrive on Enterprise, and audit logs are immutable, granular to individual actions and CSV-exportable. Custom retention, the IP allowlist and the audit log itself are Enterprise-gated, SOC 2 is listed as pending, certification scope and legal hold are not documented, and the archive leaves as a member's CSV export or return-and-delete at contract end.

Otter.ai

There are good bones — central admin controls, audit logging, SSO and SCIM on Enterprise, bulk export, and a trash folder whose permanent deletion leaves no record — but the documented default is that Otter trains its proprietary AI on audio recordings and transcriptions, escaped only by an opt-out you must request from your account manager, and aggregated data use continues regardless. We found no public information on configurable retention periods, discarding audio after transcription, legal hold, or ISO 27001 and SOC 2 certifications; the only stated security marks are TLS and AES-256.

Jamie

A German contracting entity registered at the Cologne court under German law, production data in EU data centres with transcripts and summaries in an EU-region database, a published data processing agreement with TOMs and a German DPO, a subprocessor list naming the speech recognition and AI inference providers with their regions, and ISO 27001. It stops short of the top rung: the transcription and inference operators are US companies running in the EU, the language model that reads the transcript is never named, and the authentication, email marketing and paywall subprocessors sit in the United States.

Otter.ai

Everything I can trace runs through the United States: a Delaware-registered company in Mountain View, storage on Amazon Web Services "based in the United States", and a subprocessor list in which every provider — AWS, Google Cloud, Anthropic, OpenAI, Stripe, Zendesk, WorkOS — sits in the US, with Standard Contractual Clauses and the Data Privacy Framework as the stated safeguards. A GDPR data-processing agreement exists in the terms and the chain is at least named honestly, but we found no public information on any EU hosting option for the conversations themselves.

Jamie

Public numbers exist for most tiers — a free plan at "€0 /mo" with ten AI notes a month and a thirty-minute meeting cap, a "€39 /mo. billed annually" plan, Team at "€33 /seat/mo. billed annually", "Custom billed annually" for Enterprise and a "Yearly -17%" toggle — while the terms state VAT-inclusive gross prices, monthly or yearly billing, minimum terms and notice periods. A buyer still cannot compute the full annual cost: what counts as a seat is never defined, monthly list prices for the paid tiers are not shown on the captured pricing page, and overage handling and storage limits are not documented.

Otter.ai

Per-tier detail is unusually good — minutes, import limits, chat queries, concurrent meetings and export formats per plan, with Pro at $8.33 /user/month and Business at $19.99 /user/month — but Enterprise is a sales call, the HIPAA add-on and RTMP live transcription carry unstated additional fees, SSO and SCIM require a minimum 100-user license, and the captured pages give different figures for Business transcription minutes, 6,000 on one page and unlimited on another. VAT treatment is unstated and fees are described as non-refundable, so a buyer cannot compute the true annual cost from the public pages alone.