RXed AI News

Your AI intelligence briefing
SCORE 5.4/1011/20 elements Audited 2026-07-19 · RXed table v1.0

Read AI

Capable cross-channel meeting intelligence with the most hostile onboarding in the category: a 1.5/5 Trustpilot full of 'malware' reviews and EU AI Act exposure on sentiment scoring make this a governance decision before a productivity one.

SMB toolProductivityAutomation & AgentsSearchFreemium
Vendor
Read AI, Inc. · www.read.ai
Origin
US — Seattle
Pricing
Free (5 meetings/month) · Pro ~€13 ($15 annual / $19.75 monthly) · Enterprise ~€19 ($22.50/$29.75) · Enterprise+ ~€26 ($29.75/$39.75) per user/month; Enterprise+ requires 5+ licenses; no refunds; video playback only from Enterprise up
Users (official only)
5M+ monthly active users, ~1M new users/month, teams at 90%+ of Fortune 500 (company-reported) (source, 2026-06-26)

Element scores

Reactive
Retrieval & Memory
Orchestration
Validation
Models
Primitives
Pr5
Prompts
Em5.5
Embeddings
Cx6.5
Context
Tr4
Tracing
Lg6
LLM
Compositions
Fc5
Function calling
Vx
Vector store
Rg6
RAG
Gr3
Guardrails
Mm6
Multimodal
Deployment
Ag5.5
Agents
Ft
Fine-tuning
Fw
Frameworks & harnesses
Ev
Evaluations
Sm
Small models
Emerging
Ma
Multi-agent
Sy
Synthetic data
Pc7
Protocols
In
Interpretability
Th
Thinking models
Tap or hover any element to see why it got that score.

Strengths

The cross-channel idea is real: meetings, email and messages connected into one searchable graph, with native Claude/GPT access in Search Copilot, a botless desktop app, mobile capture, and an MCP server + Claude connector + ChatGPT app shipped ahead of most competitors. Free tier gives 5 meetings/month with unlimited enterprise search, and paid starts at a competitive $15/user. $81M raised at a $450M valuation, 5M+ MAU company-reported.

Honest dings

Trustpilot sits at ~1.5/5 with a consistent pattern reviewers call 'malware': viewing a shared meeting report pushes non-users through an OAuth flow that grants calendar access and auto-joins their future meetings, and the University of Washington blocked the tool over institutional data risk. The February 2026 privacy policy openly acknowledges 'selling' identifiers to ad partners, retains audio/video up to 2 years on US servers, and performs facial-expression analysis on all participants (disabled for EU/UK). EU AI Act emotion-recognition prohibitions sit directly on the sentiment-scoring feature set. Deletion requests are widely reported as slow and frustrating, and the consent/governance controls that would fix most of this are locked to Enterprise+.

Best for: US-based individuals and sales teams who voluntarily adopt it, want meeting notes connected to CRM and email, and accept US data residency. EU organizations, regulated industries and anyone whose counterparties did not consent to recording should treat it as a compliance risk first — the intentional-user experience is decent, the bystander experience is the problem.

Sources

Every audit lists the research it rests on — transparency and traceability are the product. Tools evolve: each audit is a snapshot of its audit date, and re-audits supersede older versions (kept below for reference).

  • read.ai/plans-pricing — Official pricing (Free 5 meetings/mo · Pro $15/$19.75 · Enterprise $22.50/$29.75 · Enterprise+ $29.75/$39.75), full feature matrix, no-refund policy, Enterprise+-only HIPAA/SAML/retention/domain-capture, native Claude/GPT in Search Copilot (accessed 2026-07-19)
  • read.ai/post/read-it-whats-new-at-read-ai----februa… — Official product update 05/02/2026: desktop app (botless), Catch Me Up, shared folders, Linear integration, Google Docs/OneNote delivery (accessed 2026-07-19)
  • geekwire.com/2024/seattle-startup-read-ai-raises-50… — Series B $50M (Smash Capital, Madrona, Goodwater, Oct 2024), Seattle HQ, founders Shim/Williams/Waldron, 100k new accounts/week at the time (accessed 2026-07-19)
  • facebook.com/valenciadigitalsummit/posts/1035532632… — VDS 2026 speaker announcement (26/06/2026) relaying company-reported figures: $81M raised, $450M valuation, 5M+ MAU, ~1M new users/month, 90%+ of Fortune 500 (accessed 2026-07-19)
  • trustpilot.com/review/read.ai — ~1.5/5 aggregate; dominant complaint pattern: uninvited meeting joins, viral signup via shared reports, impersonating summary emails, deletion friction (accessed 2026-07-19)
  • cyberunit.com/insights/ai-meeting-recorder-privacy-… — Security firm's privacy-policy review (07/04/2026, policy dated Feb 2026): 'sells' identifiers to ad partners, facial/behavioral profiling of all participants, US storage, A/V retained up to 2 years, rated High risk (accessed 2026-07-19)
  • agderinthe.cloud/2026/01/19/read-ai-the-meeting-ass… — IT-admin field analysis (19/01/2026) of the viral OAuth onboarding chain and M365 tenant containment steps (Purview, Defender, Entra ID) (accessed 2026-07-19)
  • thebusinessdive.com/read-ai-review — Hands-on test (29/04/2026, scored 3.1/5): transcription/summaries solid, file upload paywalled even in Enterprise trial, privacy concerns confirmed in user research (accessed 2026-07-19)
  • basilai.app/articles/2026-05-18-eu-ai-act-emotion-r… — EU AI Act analysis (18/05/2026): emotion-recognition prohibition covers Read's sentiment scoring; Read's own EU/UK Read Score carve-out acknowledges the exposure; deployer liability for organizations using it (accessed 2026-07-19)
  • tldv.io/blog/read-ai-review-alternatives — Competitor hands-on review (updated Apr 2026, bias noted): 2.2/5 average across 123 reviews on major sites, 84% 1-star on Trustpilot, tone-deaf sentiment analysis, individual licensing complications (accessed 2026-07-19)