AgentConnect today announced the release of its open-source platform designed to enable teams and multiple AI agents to collaborate across the communication and development tools they already use. The platform addresses a growing challenge in the AI agent space: while individual agents are becoming more capable, their work often remains siloed within a single user's terminal, leading to fragmented context and duplicated integration efforts.
The new platform allows teams to connect ACP-compatible agents such as Claude Code and Codex, assign each agent a distinct role, and facilitate collaboration through shared conversations and agent-to-agent calls. This approach moves AI agents from solitary tools to integrated team members, capable of working alongside humans in familiar environments like Slack, Telegram, and Discord.
For example, during an incident response, a team can deploy a triage agent to inspect available context, a code specialist to trace a suspected regression, and a verification agent to check proposed results. Human teammates can add context, redirect work, and follow the entire diagnosis and decision-making process within the same Slack thread, enhancing transparency and coordination.
The platform's versatility extends beyond incident response. Customer support issues can transition from a public-facing Telegram conversation to a trusted Slack workspace without manual context copying. Engineering teams can combine general and specialist reviewers to assess pull request risks, with inline GitHub comments and independent check results. Scheduled and webhook-driven workflows can also continue in shared conversations when human input is required.
Key features of AgentConnect include the ability to configure multiple agents with distinct models, workspaces, memory, MCP servers, skills, repository access, and sandbox policies. The platform integrates seamlessly with existing tools, allowing work to be initiated from conversations, GitHub pull requests, webhooks, or schedules. Fine-grained access control manages agent permissions separately from session visibility, ensuring that each agent can only access necessary resources. Additionally, the platform supports linking multiple social accounts to a single member profile for consistent permission enforcement.
AgentConnect is licensed under Apache 2.0 and can be self-hosted using Docker Compose or Kubernetes, giving teams full control over their data and execution environment. The open-source nature of the project encourages community contributions and customization. The platform is available today, with source code on GitHub, and more information can be found at agentconnect.md and in the documentation.
This release signifies a shift towards more collaborative human-AI workflows, addressing the fragmentation that has hindered the scalability of AI agents in team environments. By integrating agents into existing communication channels and providing robust access controls, AgentConnect offers a practical solution for organizations seeking to leverage multiple AI agents effectively.
