Dust
Lets a company plug an AI assistant into its own Slack, Notion, Drive and GitHub so employees can ask questions and get answers pulled from actual internal documents, not the open internet.
🔗 Visit Dust
Description
A lot of AI assistants are impressive in a demo and useless at work, because they don't know anything about your company — your product docs, your Slack history, your internal wiki. Dust exists to fix exactly that gap: it connects an AI assistant directly to the tools your team already uses, so it can actually answer "what's our refund policy" or "summarize last week's engineering standup" instead of making something up.
Dust is an AI agent platform built by a team of former OpenAI and Stripe engineers. It connects large language models (from OpenAI, Anthropic, Google, or Mistral) to a company's internal data sources — Notion, Slack, Google Drive, GitHub and more — and lets non-technical teams build custom AI assistants without writing code. Agents can search across connected sources to answer questions or complete tasks, and the whole thing runs in a collaborative workspace so teams can build and share assistants together. It's SOC 2 compliant, with enterprise-grade data security, and claims over 300,000 deployed agents across 3,000+ organizations — this isn't an early-stage experiment, it's already in production use at scale.
💬 Our review
The short version: if your team's actual problem is "our AI doesn't know our own company's information," Dust solves that specific problem well and is worth the per-seat cost; if you just want a general chatbot, you don't need it.
Dust's real advantage over building this yourself with raw OpenAI/Anthropic APIs is the connector layer — Notion, Slack, Drive, and GitHub integrations that would otherwise take an engineering team weeks to build and maintain. Backed by a $16M Series A from Sequoia Capital and already past $1M in annual recurring revenue, it's not a fly-by-night tool, and multi-model support means you're not locked into one AI provider's pricing or quality swings. The trade-off is cost: per-seat pricing (roughly $29-30/month, more for Business tier) adds up fast for larger teams, and companies with strict data-residency requirements should check Dust's specific compliance posture against a self-hosted alternative before committing. Best fit: mid-size to large teams that already live in Slack/Notion/Drive and want AI assistants grounded in that data without an engineering project. Weaker fit: a solo founder or tiny team better served by a free ChatGPT plan, or a regulated enterprise that specifically needs on-premise deployment (StackAI covers that case).
💰 Pricing
📊 Global score
🤖 AI-enriched data
Gratuit disponible. Pro ~29-30 $/siège/mois (24 $ facturé annuellement). Business ~99 $/mois. Enterprise sur devis.
Pros
Connecteurs prêts à l'emploi vers Notion, Slack, Drive, GitHub — évite un projet d'ingénierie interne
Multi-modèle (OpenAI, Anthropic, Google, Mistral) — pas de dépendance à un seul fournisseur
Conforme SOC 2, déjà utilisé à grande échelle (300 000+ agents, 3000+ organisations)
Financé par Sequoia Capital, déjà rentable en ARR — pas un projet expérimental
Cons
Tarification par siège qui monte vite pour les grandes équipes
Pas de déploiement on-premise natif — vérifier la conformité pour les secteurs très réglementés
Business/Enterprise nécessitent un budget conséquent pour du volume
Peu d'intérêt pour un usage solo ou occasionnel face à un abonnement ChatGPT classique