NanoClaw wants creating an AI team to feel as simple as sending a Slack message. The Slack integration from NanoCo., the company behind NanoClaw, lets users create persistent agents with individual identities, roles, skills, instructions, memory, and permissions.
Instead of working with one general-purpose assistant, a user can build a group of agents with different jobs and toolsets. A single Slack prompt can create agents with specialized workflows and custom avatars, giving each one a defined place in the team.
From One Slack Agent to an Entire Team
The setup begins by cloning the NanoClaw project and running its nanoclaw.sh installer. The process takes a user from a fresh machine to a named agent that can receive messages, with Slack included among the supported channels.
NanoClaw’s Slack flow also removes several steps from building a traditional Slack bot. Users choose the Connect Slack option instead of manually creating apps and collecting secrets. After authorization, the first agent appears in Slack and can communicate with the user.
The first authorization connects a workspace one time. NanoClaw can then provision additional agents as Slack bots, each with its own identity. Those agents continue running on the user’s infrastructure and connect to Slack through Socket Mode.
NanoCo. says it does not store the agents’ Slack tokens. The tokens remain on the user’s machine, while organizations can apply their normal app-approval policies to the NanoClaw integration. The app’s Home tab shows provisioned agents and lets users revoke or disconnect agents or workspaces.
Agents With Jobs, Memory, and Permission Rules
The lead agent has a Model Context Protocol tool for creating new agents and defining their instructions, personas, skills, and tools. A user can ask a Slack agent to create a product manager, architect, implementation agent, code reviewer, and testing agent, with each one receiving a different set of tools.
Those agents can work together with people inside Slack channels and shared Slack Canvases. They can introduce themselves, coordinate work, and respond when tagged. Comments in a Canvas can also route back to the agent responsible for the related work.
NanoClaw agents can receive browser and internet access, memory, coding capabilities, and other tools. They can also communicate outside Slack through platforms such as Telegram or WhatsApp, which gives the same agent team more than one place to work.
Each person can have a persistent agent of their own. These personal agents can communicate with other agents under policies defined by humans, connecting individual assistants to a larger team without removing human control over their roles and permissions.
Slack Becomes the Meeting Place
Gavriel Cohen, NanoCo.’s CEO and co-founder, described the direction in direct terms: “I think this is agents arriving natively in Slack for the first time.” The design places agent creation, communication, and coordination inside a workplace that already supports channels and shared workspaces.
Slack announced in April that users could add external AI agents directly to the platform, with OAuth and environment setup handled through the platform. NanoClaw is named alongside Lovable, Hyperagent, Superhuman, n8n, Vercel, ChatGPT, LangChain, Runlayer, and Skydive on Slack’s Slack Code page.
The larger shift is from asking one assistant for an answer to managing a group of agents with separate responsibilities. Developers and enterprises can choose which underlying large language model powers their NanoClaw agents, while each agent keeps its own instructions, memory context, tools, and permissions.
Cohen put a timeline on that change: “In the next 12 to 18 months, everyone on a team will be a manager of agents.” NanoClaw’s Slack integration is built around that idea, turning agent creation into a workspace task and giving each new assistant an identity that can persist across conversations and assignments.
Based on
