RXed AI News

AI to the bone.
Audited 2026-08-08 · RXed table v1.0

Dust

Visit dust.tt
“Multiplayer AI for human-agent collaboration” — the vendor’s own words

The fastest way to wire agents into Slack, Drive and Notion for €29 a seat — until the 1GB/user data cap and Enterprise-only governance push you into a custom quote.

The verdict, read by Jan · AI clone of Jan’s voice, generated locally with his consent.
Best for: Slack-centric teams and mid-to-large enterprises that want employees building and running their own agents against company knowledge without waiting on an engineering team to ship each one.
Scope17/20
Quality6/10
Where the quality sits
7Reactive
5Retrieval & Memory
7Orchestration
6Validation
6Models
Enterprise platformAutomation & AgentsProductivitySearchFreemium
Vendor
Dust · dust.tt
Origin
France — Paris
Pricing
Trial Free · Pro ~$31/user/mo · Enterprise Custom
Users (official only)
300,000+ agents deployed, 3,000+ organizations running on Dust (self-reported live homepage stats); funding coverage of the May 2026 Series B additionally cites ~70% weekly active usage and zero churn in 2025 (source, 2026-08-08)
TrialFree15-day trial of Pro features, no permanent free tier
Pro~$31/user/mo€29/user/month excl. tax; 1GB/user data cap
EnterpriseCustom100+ users; SSO, SCIM, regional hosting

No perpetual free tier — the 15-day trial is the only free access. The €29/user/month Pro price and 1GB/user data cap are independently corroborated by three separate reviewer sources, not just the vendor.

checked 2026-08-08 · vendor pricing page

Element scores

Reactive
Retrieval & Memory
Orchestration
Validation
Models
Primitives
Compositions
Deployment
Emerging
Tap or hover any element to see why it got that score.

Strengths

A well-documented RAG pipeline (semantic search, document retrieval, a genuine 500k-token Extract Data action) plus real enterprise security posture shipped by default: SOC2 Type II, GDPR compliance, zero training on customer data, RBAC with dual-layer agent permissions, 365-day audit logs and a 99.9% uptime SLA. Multi-model choice across GPT, Claude, Gemini and Mistral, MCP support for proprietary systems, and traction that's hard to fake — 300,000+ agents deployed across 3,000+ organizations with zero reported churn in 2025.

Honest dings

No dedicated evaluation or regression-testing product, no fine-tuning, no synthetic data, no interpretability tooling. The Pro tier caps data sources at 1GB per user — reviewers consistently flag this as the point where real deployments need Enterprise, which is custom-priced and requires 100+ seats.

Prices and details change — this passport is re-verified at least quarterly.
Sources (12) — every claim traceable

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).