Dust
AI agent platform for building and deploying LLM-powered workflows and assistants.
Dust doubles down on MCP-native agents with multi-model routing and enterprise guardrails.
◆Recent moves
- 1mo ago
Compact mobile input bar goes voice-first
A redesigned mobile input bar centered on voice. Fits the trajectory of Dust as a daily-driver agent surface rather than desktop-only, and hints at a deeper voice-mode push to follow.
- 1mo ago
Frames export as PDF or PNG
Agent-produced frames can now leave Dust as PDF or PNG, closing the loop for content-generation workflows. Pairs with the inline frame editing shipped the same day — together they treat frames as a real authoring surface rather than disposable LLM output.
- 1mo ago
Inline text editing in frames
Users can tweak frame content directly instead of reprompting. Combined with the same-day PDF/PNG export, it signals Dust treating frames as an authored deliverable, not just an agent transcript.
- 1mo ago
Gemini 3.5 Flash now available
Another model wired into the routing layer the day after a Grok and Anthropic refresh. Reinforces model neutrality as a structural position rather than a marketing line.
- 1mo ago
Grok 4.3 upgrade and Anthropic model migrations
A coordinated multi-vendor model refresh — Grok bumped, Anthropic models migrated. With Gemini 3.5 Flash following the next day, Dust is making model-router cadence a visible competency.
- 1mo ago
Asana MCP update
An action-required migration of the Asana connector to MCP V2. Tactical for Asana users, but the V2 move signals where Dust expects the broader MCP integration surface to land.