Most AI coding tools are assistants: they suggest, you type. Lume v2 takes the next step. With SOLO, our autonomous coding agent, the workflow inverts — you describe the outcome you need, the agent plans and executes the work, and you review the result before it ships.
What SOLO does
SOLO takes a task — a feature, a fix, a refactor — and drives it end to end: it gathers the right context from your codebase, picks its tools, writes and tests the code, and returns a complete, reviewable solution. It is now available in beta on desktop, web and mobile.
Crucially, SOLO is not a separate product to learn. Lume IDE and SOLO share one platform, so teams can switch between hands-on editing and autonomous delivery in the same project — assign routine work to the agent, keep the interesting problems for yourself.
Built for teams that ship
Under the hood, Lume runs multiple agents that cooperate on harder problems, integrates external tools and data through the Model Context Protocol, and predicts your next edit as you type. More context means more accuracy — especially in bigger codebases.
Privacy is a design decision, not a checkbox: Lume is local-first with minimal data collection, offers regional deployment, and is SOC 2 certified.
Why it matters
Software teams everywhere face the same equation: demand for software grows faster than engineering capacity. Autonomous agents change the economics of shipping — a single engineer directing agents can deliver what used to take a team. Lume is Summit's answer to that shift, and SOLO is its engine.