Sistava

Dust.tt

No-code platform for building custom AI agents over company knowledge and tools

About Dust.tt

Dust.tt is a Paris-based enterprise AI platform that lets teams build custom agents on top of company knowledge and connected apps. The product is designed around no-code agent creation: you wire Notion, Google Drive, Slack, GitHub, Salesforce, Zendesk, and Snowflake into a workspace, then configure agents that answer questions, draft documents, or run actions through Dust Apps and the API. It supports OpenAI, Anthropic, Google, Mistral, and DeepSeek models, with SSO, SCIM, SOC 2, and optional zero data retention for enterprise buyers. Dust earned its reputation with technical and ops teams at companies like Alan, Qonto, and PennyLane. The shift in 2026 is what they call multiplayer AI: shared agent workspaces where humans and agents collaborate on the same task. It is a strong fit for organizations that want to own their agent stack and have engineering capacity to maintain it. It is a thinner fit for solo founders or small teams who would rather hire a pre-built role than configure one from scratch.

Platform details

Official website

What does Dust.tt actually do?

Dust.tt is a no-code platform for building custom AI agents that read your company knowledge and act through connected tools. A team sets up a workspace, plugs in sources like Notion, Google Drive, Slack, GitHub, Salesforce, and Snowflake, then configures agents that answer questions, draft documents, or trigger actions through Dust Apps and the API. The platform supports OpenAI, Anthropic, Google, Mistral, and DeepSeek models, and ships with SSO, SCIM, SOC 2, and optional zero data retention. The original wedge was a great chat interface over company knowledge. The product has grown into a broader agent platform with around 50 native connectors, a visual agent builder, and a developer-friendly Dust Apps layer for custom logic. Enterprise buyers also get governance features like audit logs, role-based controls, and US or EU data residency. Where it stops short for solo founders: Dust gives you a toolbox to build agents, not pre-trained roles. You still decide who the agent is, what it does, what tone it uses, and how it coordinates with the rest of your workflow. Sistava lives at the other end of that spectrum, where you hire Maya the marketer or Tom the salesperson and the role is defined for you on day one. Net-net, Dust is a strong fit if you already have engineering or ops talent who wants a canvas. Sistava is the better fit if you want a working team this week and you would rather skip the configuration stage.

How much does Dust.tt cost?

Dust overhauled its pricing in mid-2026, replacing the old flat 29 EUR per seat, unlimited fair-use-messages model with a credits-metered structure. Pro now runs $30 per seat per month ($24 billed annually) with 8,000 credits per seat, and a new Max seat sits at $150 per month ($120 annually) with 40,000 credits, aimed at heavy users who outgrew Pro. Enterprise pricing is custom, requires a minimum of 100 members, and is quoted on an annual contract that includes SSO, SCIM, dedicated account management, and US or EU data residency. There is no permanently free tier, only the 14-day trial. Credits blend token usage and action usage, so a seat that mostly reads documents stretches further than one that calls tools on every message. Pro includes model choice across Claude, GPT, and Mistral, around 1GB of storage per user, and one private workspace per seat. Enterprise adds expanded storage, advanced security controls, priority feature access, and premium support. For a solo founder running seven AI roles, Dust Pro would charge for seven seats, which lands at roughly $210 per month before anyone runs past their credit allowance. The same coverage on Sistava is one flat plan, because pricing is per workspace and not per employee. That gap widens fast once you add specialists or a heavier Max seat. Sistava starts with a free tier and the Personal plan covers most solo-founder use cases. The full plan ladder lives at our pricing page, where every tier is self-serve and there is no minimum-seat requirement to reach the higher capabilities.

When does Dust.tt beat Sistava?

Dust wins when an organization already has an engineering or RevOps team that wants a flexible canvas to build internal agents. Wiring Notion, Slack, GitHub, and a CRM into a custom assistant is what Dust was designed for, and the Dust Apps layer is genuinely good for teams who can code. It also wins on document-grounded chat at enterprise scale. The connector library is mature, the model choice is broad, and the enterprise governance story (SSO, SCIM, SOC 2, US or EU residency) is well-tuned for 100+ seat buyers who care about audit and compliance. If your goal is a smart internal assistant that lives inside the existing knowledge stack and supports many specialists, Dust is a credible choice. Their recent 40m Series B from Sequoia and Abstract is funding more multiplayer-AI features that lean further into shared workspaces. Sistava is built for a different buyer: a founder or operator who wants pre-built roles, a team structure, sprints with a goal, and execution surfaces (browser and desktop) without configuring any of it. Different product, different north star.

Where does Dust.tt fall short for solo founders?

Dust is a builder, not a roster. The first hour on Dust is spent picking sources, configuring the agent, choosing a model, and writing instructions. The first hour on Sistava is spent meeting an AI employee, giving them a goal, and watching them get to work. For a solo founder with no spare hours, the gap matters. There is no team structure. Each Dust agent is an independent assistant. Sistava employees report into team leaders, run sprints, hold weekly planning meetings, and delegate work to each other. That changes what you can ask for: on Sistava you can request a campaign and watch the marketing team split the work; on Dust you would coordinate the steps yourself. Execution surface is narrower. Dust shines at reading documents and calling APIs. Sistava employees also run a desktop companion and browser, so they can use any web app a human can use, even ones without an API. That matters when half your stack is built on tools that never shipped a clean integration. Per-seat pricing penalizes range. Adding a third or fourth specialist on Dust adds another 29 EUR per month each. On Sistava, the price is set by the workspace, so a one-person company can run a marketing, sales, support, and ops team without a per-head tax.

Comparison