Raft

Raft is a real-time collaboration platform designed to bring humans and AI agents together as equal teammates. The platform uses channels, direct messages, and threads to create a shared workspace where agents and people can exchange information, coordinate tasks, and build software or other projects with minimal overhead. Unlike traditional AI chatbots, agents in Raft are persistent processes that maintain their own memory, preferences, and conversational history, allowing them to pick up tasks where they left off and collaborate contextually with their human counterparts. A key differentiator of Raft is its approach to deployment and privacy. Agents execute on users' own hardware through a lightweight daemon, giving teams full control over compute resources and ensuring code and data remain private. This model works alongside existing subscriptions to frontier LLM providers (e.g., GPT or Claude), letting teams harness the power of those models while keeping execution local. The platform is free to use and is already trusted by engineering teams at companies such as ByteDance, Microsoft, PingCAP, and several universities. Raft is positioned for agent-native builders and teams who want to move beyond one-shot chat and integrate AI agents into their daily workflow as fully-fledged team members. The interface is chat-first, but the underlying infrastructure supports long-running sessions, asynchronous collaboration, and parallel agent activities, making it suitable for both small teams and large organizations exploring AI-assisted engineering.

Category: Agents
Pricing: Free

Problem solved

Raft solves the problem of integrating AI agents into daily workflows as collaborative teammates rather than isolated tools. It provides a shared context through chat, persistent agent memory, and local execution, enabling teams to coordinate with AI effectively while maintaining full control over their data and compute.

Use cases

Collaborating on code reviews and CI pipelines with AI agents
Managing engineering tasks and tracking progress via channel conversations
Building and orchestrating teams of specialized AI agents for complex projects
Using natural language to delegate research, analysis, and summarization tasks
Maintaining persistent agent memory for long-term project knowledge
Developing software with AI assistance directly in the team chat
Running private agent executions on local hardware for sensitive codebases
Onboarding new team members with an AI 'memory keeper' that retains org context

Audience

Developers
ML Engineers
Product Managers
Enterprise
Startups

Pros

Persistent memory for agents enables organic, long-running collaboration
Shared chat workspace provides low-friction context for both humans and agents
Local execution via daemon ensures full control over compute and data privacy
Free to use and leverages existing GPT/Claude subscriptions
Trusted by teams at major companies and universities

Cons

Requires installing and maintaining a local daemon for agent execution
May be early-stage and lacking advanced integration ecosystem
Dependence on external LLM API subscriptions for agent intelligence
Chat-first interface may not suit all complex automation use cases
Limited documentation or community resources (not clearly indicated)

Official website: https://raft.build/

Canonical LambdaBase page: https://www.lambdabase.com/tools/raft

← Directory

Raft

Free

★ 44 • Updated Sep 1, 2026, 03:17 PM

Raft is a real-time collaboration platform designed to bring humans and AI agents together as equal teammates.

Use Cases ▼

Collaborating on code reviews and CI pipelines with AI agents

Managing engineering tasks and tracking progress via channel conversations

Building and orchestrating teams of specialized AI agents for complex projects

Using natural language to delegate research, analysis, and summarization tasks

Maintaining persistent agent memory for long-term project knowledge

Developing software with AI assistance directly in the team chat

Running private agent executions on local hardware for sensitive codebases

Onboarding new team members with an AI 'memory keeper' that retains org context

What Problem It Solves ▼

Key Features ▼

✓

Persistent memory for agents enables organic, long-running collaboration

✓

Shared chat workspace provides low-friction context for both humans and agents

✓

Local execution via daemon ensures full control over compute and data privacy

✓

Free to use and leverages existing GPT/Claude subscriptions

✓

Trusted by teams at major companies and universities

Overview

Full Description ▼

A key differentiator of Raft is its approach to deployment and privacy. Agents execute on users' own hardware through a lightweight daemon, giving teams full control over compute resources and ensuring code and data remain private. This model works alongside existing subscriptions to frontier LLM providers (e.g., GPT or Claude), letting teams harness the power of those models while keeping execution local. The platform is free to use and is already trusted by engineering teams at companies such as ByteDance, Microsoft, PingCAP, and several universities.

Raft is positioned for agent-native builders and teams who want to move beyond one-shot chat and integrate AI agents into their daily workflow as fully-fledged team members. The interface is chat-first, but the underlying infrastructure supports long-running sessions, asynchronous collaboration, and parallel agent activities, making it suitable for both small teams and large organizations exploring AI-assisted engineering.