NanoCo added an integration between the open-source agent platform NanoClaw and Slack, allowing new artificial intelligence agents to be created and specialized teams to be formed from them through a single message inside the workspace. Each agent receives an independent identity in Slack, including a name, avatar, and separate account, along with a customizable role, memory context, instructions, permissions, and tools.
The integration arrives at a time when several platforms are competing to bring artificial intelligence agents into collaboration tools. However, NanoClaw focuses on enabling an existing agent to create additional digital colleagues during a conversation, rather than building every agent in advance through a separate management tool. NanoCo describes this approach as the first of its kind in Slack, a marketing claim originating with the company itself.
Initial connection, then creation of additional agents
The setup process begins by installing NanoClaw on a local device or cloud virtual machine, then running the nanoclaw.sh installation file, which sets up dependencies, credentials, the agent container, and the connection to the first messaging channel. After selecting Connect Slack and authenticating through Slack, the user adds NanoClaw to the workspace via Add to Slack.
The NanoClaw listing in the Slack Marketplace says that the workspace is connected once, after which NanoClaw can provision additional agents as separate Slack bots. These agents run on the customer’s own infrastructure and connect to Slack via Socket Mode. NanoCo also says it does not store Slack tokens, as the tokens remain on the user’s device, while Slack’s usual administrative approval policies remain in effect.
The app’s Home tab displays the agents provisioned inside the workspace, with the ability to revoke an individual agent or disconnect the entire workspace. Therefore, the integration does not eliminate the need to run and manage NanoClaw infrastructure; instead, it serves as a one-time bridge between that infrastructure and Slack.
Specialized teams working inside channels and Canvas
According to Gavriel Cohen, CEO and co-founder of NanoCo, the primary agent can use a tool built on the Model Context Protocol (MCP) to create agents and define their personalities, skills, and tools, then place them in shared rooms. A user can request one agent to review code, another to test, and a third to write or manage products, allowing them to exchange tasks and enabling one to ask another for clarification when needed.
The agents support working with Slack Canvas and communicating with users and peers inside the channel. In a marketing demonstration, a primary agent named Nano created agents called Atlas for strategy, Sage for content, Echo for social media, Scout for outreach, and Compass for search engine optimization and analytics. NanoCo says the agents typically respond only when mentioned, which is intended to reduce the risk of them triggering one another indefinitely.
What distinguishes NanoClaw in practice?
NanoClaw differs from Claude Tag, ChatGPT Workspace Agents, and Salesforce Agentforce in one fundamental respect: new agents can be created from within a conversation with an existing agent, with each agent receiving an independent Slack identity. The other services generally require an administrator or user to create the agent first through a dedicated tool and then connect it to channels.
By contrast, NanoClaw shifts more control and responsibility to the organization. The platform is open source and customizable, and users can choose the large language model used, including Claude Code by default, Codex, OpenCode, and local Ollama models, according to the cited documentation. However, the organization also bears responsibility for managing permissions and memory, inference and hosting costs, and monitoring failures.
Context and availability
NanoClaw was released under the MIT License on January 31, 2026, as a smaller alternative to OpenClaw with a greater focus on security isolation. In March, the project partnered with Docker to run agents inside Docker Sandboxes, and NanoClaw 2.0 then added Vercel’s Chat SDK and OneCLI’s dependency gateway in April. In May, Gavriel Cohen and his brother Lazer Cohen founded NanoCo and raised a $12 million seed round led by Valley Capital Partners, with participation from Docker, Vercel, monday.com, and others.
The integration is available through the Slack Marketplace, and NanoCo does not charge fees for Slack’s community capabilities, according to Cohen’s statements, but users may incur model and hosting expenses. Agents can also access Slack, Telegram, and WhatsApp while retaining shared workspace and memory, although conversations remain separate sessions.