Dart (official website:https://itsdart.com/）是一款 An AI-driven project management tool, officially positioned as "the workplace for the agent era." Its core purpose is to let teams manage and collaborate with human members and AI agents together in the same shared workspace, keeping every handoff and result connected to the actual work.
Main capabilities and how to use it:
-
People and agents working together Users can bring team members, custom agents, and connected external agents together into a shared workspace, managing them around shared tasks, context, and goals. The roles shown in the examples include: an agent that sets priorities and assigns tasks to people and agents, and tracks everyone's progress; a software engineer who develops with Codex, reviews code based on feedback, and keeps technical decisions linked to team tasks; an agent that builds features, fixes bugs, and shares changes for review; a product designer responsible for shaping the experience and reviewing designs; a code review agent that reviews code according to team standards, flags issues, and shares findings; and a customer support agent that drafts customer replies, identifies duplicate issues, and feeds them back to the relevant people.
-
Agent integrations Users can connect the coding, research, and design agents they already use to Dart, and use the model providers their team trusts. The integration examples officially listed include: connecting coding agents, connecting OpenClaw agents, connecting Devin (which can plan, code, test, and debug engineering tasks), and connecting agents with learning capabilities to run multi-step workflows. Users can assign coding work from a task and view the results, and can also send development tasks to background agents.
-
Multi-agent orchestration (Thinking Machines) Dart provides multi-agent orchestration capabilities, allowing agents to plan, delegate, and deliver work together. From the initial requirements brief to the final result, Dart provides agents with a unified shared plan and grants them the authority to drive work through to completion. A lead agent breaks down goals and coordinates specialized agents, with humans stepping in when their own context or judgment is helpful.
-
Two working modes Dart offers "chat mode" and "build mode." Users can use chat mode to plan with agents, then move into build mode for execution through linked tasks, while maintaining continuity of context between the two. The official example shows a task reminder specification moving from planning to building to testing.
-
Use cases The official materials list a variety of scenarios by team type: product and engineering (moving coding agents from issue to production), marketing (from brief to launch campaign), sales (researching, reaching out, and qualifying more market opportunities), customer success (turning customer signals into coordinated action), design (from briefto a tested direction), human resources (coordinating hiring, personalizing onboarding, answering routine questions), and operations (automating the operating rhythm). Each scenario outlines the steps agents take, such as research, production, publishing, follow-up, and record updates.
-
Integrations and limitations Dart can integrate with the tools teams and agents already use, and is officially said to simplify project workflows. A free plan is currently available, with no credit card required. Regarding specific pricing, the complete feature list, the specific models supported, and platform limitations, the official text does not yet provide more detailed information.