AI BriefingLemma Releases Open-Source AI Agent Workspace Built Primarily by Claude
Key point
The platform allows agents to run on schedules or triggers with shared memory and requires human approval for risky actions.
Details
Lemma has released an open-source workspace where humans and AI agents collaborate, with the majority of its code written by Claude and some agents running on DeepSeek. The development process involves humans defining plans that agents execute directly within the Lemma environment, creating a self-building product loop.
Core Functionality
Agents in Lemma operate on schedules, row changes, or webhooks, utilizing persistent browser sessions and shared team memory. Results are stored in tables, with a mandatory human approval step for any risky actions. Users can interact with agents via Slack, Telegram, WhatsApp, Teams, or email.
Deployment and Access
The platform is available under the AGPLv3 license and supports self-hosting via Git and Docker. A hosted version with paid usage is also available. Users can define workspaces in plain English using Claude Code, which generates the necessary tables, agents, schedules, and web apps through Lemma's CLI, leveraging existing Claude logins without requiring separate API keys.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
DeepSeek Launches Open-Source Harness Desktop for macOS and WindowsShow HN: Audionaut – Open-Source Multitrack Audio Editor with MCP Support for AI Agents