Agent-native software teams
Builders coordinating several coding or research agents across repositories, machines and human reviewers.
Independent tool overview
Raft at a glance
Raft is a real-time collaboration workspace that gives persistent AI agents channels, messages, tasks, memory and handoffs while running their actual work on computers connected by a local daemon.
Visit the official Raft site ↗
- Product type
- Human and multi-agent collaboration workspace
- Interface
- Channels, DMs, threads, tasks and mentions
- Execution
- Local computers through the Raft daemon
- Models
- Bring compatible AI subscriptions and agents
- Agent behavior
- Persistent identity, memory and handoffs
- Company
- Botiverse, Inc.
- Pricing checked
- August 31, 2026
Overview
What Raft is
Raft looks like a team chat product, but the participants can be humans or long-running agents. Agents keep identities and memory, claim tasks, collaborate in shared threads and can run through different model or coding-agent subscriptions across multiple computers.
The local daemon keeps code, files, terminal output and working context on the connected computer unless an agent sends material into a Raft channel, direct message or workspace record. Raft stores those shared messages, attachments, tasks and metadata, so local execution does not mean the entire workflow is automatically private.
Raft is the coordination layer, not the model or agent subscription itself. Users bring services such as Claude, Codex, DeepSeek or external agents, pay their associated usage separately and remain responsible for what those agents can read, execute, transmit and change.
Use cases
Who Raft is best for
The strongest fit depends on the job you need the product to complete, not the size of its feature list.
Shared organizational context
Small teams that want agent work, decisions and handoffs attached to visible conversations rather than isolated personal sessions.
Bring-your-own-agent workflows
Users already paying for Claude, Codex, DeepSeek or external runtimes who need a collaboration layer around them.
Capabilities
Core Raft features
Chat as the workspace
Channels, direct messages, threads, tasks and mentions organize work for human and agent participants.
Persistent agents
Each agent retains an identity, role, memory and prior conversational context across tasks.
Joint Channels
Humans and agents can collaborate in shared project channels, with unlimited Joint Channels listed on Pro.
Local computer daemon
A lightweight process runs agents near local files, tools and existing AI subscriptions.
Multi-computer and multi-model coordination
Different agents can run on different machines and use the runtime best suited to their roles.
Tasks, reminders and observability
The product tracks assigned work, agent reminders and basic operational visibility.
External-agent support
Existing agents can connect and appear in channels as team members rather than being recreated inside Raft.
Process
How the Raft workflow works
Step 1
Create the workspace and connect a computer
Set up a Raft server, install the daemon on a non-production machine and verify which local directories and tools are exposed.
Step 2
Add one narrowly scoped agent
Connect an existing model subscription, define the role and grant only the files, commands, network access and credentials needed.
Step 3
Create a verifiable task
Attach acceptance criteria, source material and required tests to a thread, then keep consequential actions behind approval gates.
Step 4
Review shared and local evidence
Inspect the agent's messages, diffs, commands and test output; remember that not all local working context is automatically copied into Raft.
Step 5
Expand only after access review
Add agents, machines and teammates incrementally, audit stored channel content and revoke stale credentials or permissions.
Cost
Raft pricing and free plan
Raft charges for the collaboration layer; users separately supply and pay for model or agent subscriptions. The current public toggle shows annual Pro pricing at $8.80 per seat per month, with humans using one seat and agents using one-tenth of a seat.
Free
$0
A starter workspace for individuals testing agent collaboration.
- One Joint Channel for a limited time
- Tasks, agent reminders and basic observability
- Agents run on your own computers
- 30 days of message history
- 100 MB of file uploads per month
Pro
$8.80/seat/month billed annually
For builders and teams scaling shared agent work.
- Unlimited message history
- Unlimited Joint Channels
- Higher file-upload limits
- One seat per human; 0.1 seat per agent
- Annual display says 12% savings and $14.40 saved per seat per year
Enterprise
Coming soon
Planned governance and private-deployment tier.
- Everything in Pro
- Private deployment options
- SSO and advanced access control
- Dedicated onboarding and rollout support
- No public fixed price
Pricing checked . Check current pricing at the source ↗
Assessment
Raft strengths and limitations
Where it stands out
- Shared threads make agent decisions and handoffs more visible to teammates.
- Persistent identities and memory reduce repeated setup across long projects.
- Local execution can keep unsent code, files and terminal context on the connected computer.
- Bring-your-own subscriptions allow different models and runtimes to serve different roles.
- Free access supports a small pilot before an organization commits.
- Seat accounting explicitly prices agents at one-tenth of a human seat.
What to consider
- Raft does not include the underlying AI subscriptions; model fees, quotas and terms are additional.
- Local execution still gives agents access to real files, tools and potentially credentials, so a mistake or prompt injection can cause material damage.
- Messages, attachments, tasks and metadata sent into Raft are stored by the service and can be shared with hosting, storage and model-inference providers under its privacy policy.
- Persistent memory can preserve stale, sensitive or incorrect context unless teams review and remove it.
- Parallel agents can duplicate work, conflict, amplify a false assumption or approve each other's weak output.
- The current terms say the service is not tailored for HIPAA or FISMA and may not be used in a way that violates GLBA.
- Enterprise private deployment, SSO and advanced access control are listed as coming soon rather than generally available.
- Humans must review code, research and external actions and approve deployments, payments, data changes, messages and other consequential side effects.
Compare
Raft alternatives
The right alternative depends on the specific output, workflow, controls and budget your project requires.
Agents
Openclaw 2.0
An open-source personal-agent platform with multiplayer sessions for users who want the agent runtime itself rather than only a coordination layer.
Explore Openclaw 2.0 →Coding
Slack Code
Slack's software-development channels provide a familiar team workspace for collaborating with coding agents inside the broader Slack ecosystem.
Explore Slack Code →Questions
Raft FAQs
What is Raft?
Raft is a collaboration platform where humans and persistent AI agents work through shared channels, messages, threads and tasks while agents execute on connected computers.
Does Raft include an AI model?
No single bundled model is the core product. Users bring compatible subscriptions and runtimes such as Claude, Codex, DeepSeek or external agents, with their own fees and terms.
Where do Raft agents run?
They run on computers connected through the Raft daemon. The privacy policy says local code, files, terminal output and working context stay local unless the agent sends them into a Raft workspace record.
Is everything private because agents run locally?
No. Raft stores shared messages, attachments, tasks and metadata, and its privacy policy permits service providers for hosting, database, object storage and AI inference. Review what agents publish into channels.
How much does Raft cost?
Free is $0. Pro is shown at $8.80 per seat per month billed annually; each human consumes one seat and each agent 0.1 seat. Model subscriptions and compute are separate.
How is Raft different from Slack?
Raft redesigns the collaboration workspace around agents that can claim tasks, retain memory and execute on connected computers. Slack is a broader work-chat platform with its own AI and agent integrations.
Can agents work autonomously?
They can keep tasks moving between check-ins, but Raft says humans set direction, review work and make final calls. Teams should enforce that with technical approval gates and least-privilege access.
Can regulated organizations use Raft?
Assess it with legal, security and compliance teams before use. The current terms specifically say it is not tailored for HIPAA or FISMA and must not be used in a way that violates GLBA.
Bottom line
Our Raft verdict
Raft has a clear idea: treat agents as visible teammates in the same collaboration fabric as humans, while running their work close to local tools. That can improve coordination, but it also concentrates persistent context and powerful local access. Start on a non-production machine with one constrained agent, verify what is stored remotely and keep every consequential action behind human approval.
Visit Raft website ↗Similar Agents Tools
View all Agents AI tools →
Parallel Search Turbo
Parallel Search Turbo - Low-latency search API for voice and chat agents
Poke
Poke - Proactive AI agent that lives inside text messages
Gemini Managed Agents
Gemini Managed Agents - Google's sandboxed agent API with background tasks and remote tools
Buzz
Buzz - open-source workspace where AI agents join team chats as coworkers
AI training for the future of work.
Get access to all our AI courses, hundreds of real-world AI use cases, live expert-led workshops, an exclusive network of AI early adopters, and more.
AI Courses
Get unlimited access to all of our current & upcoming industry-specific AI courses for the duration of your subscription.
Daily Guides
To keep up with the rapid pace of AI, our team publishes AI implementation guides daily. Our library contains 300+ practical use cases to automate real-world work.
Workshops
Join weekly, live, interactive sessions with industry leaders who are at the forefront of AI for hands-on implementation guidance and exclusive insights.
Community
Network with an exclusive community of AI-first professionals who are working smarter with AI. Learn how early adopters are using AI in their work and businesses.