The workspace built for agents, humans and AI-native companies

Make every AI agent work like it knows your company.

Driftless is the workspace that keeps your team's decisions, projects, documents, and operational knowledge connected, and ready for every agent.

Driftless topic graph — 257 topics and 396 relations, the team's living context

Connected, not copied

Built on the systems your work already lives in.

GitHubGitHub
SlackSlack
NotionNotion
DriveDrive
LinearLinear
SalesforceSalesforce
GmailGmail
GitHubGitHub
SlackSlack
NotionNotion
DriveDrive
LinearLinear
SalesforceSalesforce
GmailGmail

One real example, start to finish

Works with your company context on Driftless at Claude.

Capture, right from the conversation

An agent proposes what it learns — anchored to the real code, submitted for review — without leaving the chat.

Claude proposing a real topic — Warm Pool, anchored to src/runtime/sandbox — sent to the inbox for review

Delivered back, with the source attached

The next agent asks, gets the current answer with its source attached, and acts on it — not a guess.

ChatGPT calling the Driftless MCP tool and retrieving the same Warm Pool context to answer a real question

Built for how each team already works

Team chats

Give Claude or ChatGPT the context they need to answer like they work here.

Coding agents

Anchor decisions and gotchas to the code your agents are about to edit.

Automations

Pull governed context into the workflows and scripts you already run.

Support & ops agents

Power customer-facing agents with context that's current and approved.

Your own product

Build directly on Driftless via CLI, MCP, or REST API.

CLIMCPREST

Not a doc dump — a governed repository

Everything AI needs to understand your company.

Everything, organized

Every topic filed under a real area — nothing floats in Unassigned.

The real Driftless context list — areas like Agents & Runtime, Growth & Content, Integrations & MCP, organized and searchable

Anchored to real code

Not a vague note — tied to the actual files it's about.

A real topic — Model Gateway — anchored to real files like apps/api/src/agent-runs/model-http.ts, not a vague note

Nothing goes silently stale

Every change, every usage, tracked over time.

The real history log for a topic — every modification, file change, and usage tracked over time

Comments

Leave feedback on a topic without touching its content — human or agent, either can leave one.

Permissions

Every request is scoped to a workspace — remove someone and their access is gone immediately.

Security

Encrypted in transit and at rest. Credentials are never stored in the clear.

A workspace on top, a platform underneath

Built for AI-native companies and companies that try to go native.

Web

The workspace your team works in every day.

CLI

Read and write context from your terminal.

MCP

Your agents reach the same system you do.

API

Build directly on the substrate underneath.

No markup on inference

Bring your own model. Keep your own context.

Connect OpenAI, OpenRouter, DeepSeek, Qwen, OpenCode, any OpenAI-compatible endpoint, or a local model. You pay your provider directly, at their price — Driftless never marks up a token.

OpenAIOpenAI
OpenRouterOpenRouter
DeepSeekDeepSeek
QwenQwen
OpenCodeOpenCode
OpenAIOpenAI
OpenRouterOpenRouter
DeepSeekDeepSeek
QwenQwen
OpenCodeOpenCode
Custom OpenAI-compatibleLocal / self-hosted

Not only context, a workspace too

Not just context — work that actually gets done.

Projects — the agent works the loop

Pulls the next card, does the work, validates it, marks it done. Not a suggestion — an executed step.

A real terminal loop — Claude Code pulling the next Project card and working it via the Driftless CLI

Collections — real pipelines, moved forward

An agent updates a real record — stage, status, notes — the same way a teammate would.

Claude moving real candidates through the Hiring collection — offer_pending, rejected — with notes

Chat, built in — not bolted on

Driftless's own chat answers from Collections and Knowledge together, with sources cited.

Driftless's native chat answering a real question, citing Collections and Knowledge topics with sources

Teams already building this way

The piece their stack was missing.

Every decision, every constraint, every mistake we already made — anchored to the code. The agent retrieves it before touching that area. No one has to explain it again.

Juan Bustamante

Driftless became our team's technical memory. It helps us capture the knowledge that usually lives in a few people's heads, keep it up to date alongside the code, and automatically catch when something's gone stale. It's one of those tools that becomes indispensable fast.

Juan Cheli

We work with 7 coupled repos and code agents across all of them. Critical knowledge like auth contracts, our multi-tenant isolation model, and the learnings from every incident used to get lost between sessions. Driftless turned all of that into persistent context the agents can query on their own. It's the piece our stack was missing.

Ian Salzgeber

Start free, grow when your team does

Stop starting from zero.

Free

For trying Driftless

$0

Everything you need to get going.

  • 1 workspace, 1 repo
  • Up to 5 projects, 5 collections
  • Your team's knowledge, always current, unlimited
  • Works inside your AI tools
  • Connect your GitHub
Start free

Pro

For solo builders

$9/mo

More space, your own models.

  • Everything in Free
  • 3 workspaces
  • Unlimited projects & collections
  • Chat + agents on your own model (BYOM)
Start free

Team

Most popular

+$15/seat, +$15/workspace beyond that

$49/mo for 5 seats

One shared workspace, governed.

  • Everything in Pro
  • 5 seats included (add more anytime)
  • Members, roles & approvals
  • Integrations: Notion, Google Drive, and more
Start free

Enterprise

Org as system of record

Custom

Your systems of record, governed.

  • Everything in Team
  • Unlimited seats & workspaces
  • SSO & audit log
  • Custom integrations, dedicated support
Talk to sales

Every plan includes unlimited knowledge, the CLI, and MCP. Bring your own model from Pro up.

Give every AI agent your company's context.

Turn what your company knows into work that gets done.