Derive Labs
Derive is a workspace for work made with AI agents: it gives HTML, Markdown, decks, documents, and built sites durable URLs, version history, access controls, and review collaboration. Its stated aim is to make working in the age of AI more seamless and fulfilling by keeping agent output findable, reviewable, and owned by the user.
At a Glance
- Individuals and independent projects
- Small teams
- Teams whose agents ship work requiring review
- Organizations with control, accountability, SSO, audit, and residency requirements
- +1 more
AI Tools by Derive Labs
(1)
Derive
AI Artifact Version Library
Discussions
No discussions yet
Be the first to start a discussion about Derive Labs
Latest News
Show HN: Derive – An open home for AI artifacts and workflows
Share a URL with an agent and it reads the document: artifact URLs now serve Markdown on request.
Derive 0.2.0 released as the first tagged release, introducing the documented artifact/review/agent workflow.
@derive-to/cli 0.5.0 published on npm for publishing, review, comments, and revision workflows.
Products & Services
Hosted workspace where users and agents publish HTML, Markdown, decks, documents, and whole built sites to durable URLs with versions, comments, reviews, sharing, access controls, and a library.
The complete product in one container, with API, web app, sign-in, publishing, comments, and sandboxed viewer; SQLite/local disk are defaults, with Postgres and S3/R2 supported at scale.
Node.js command-line client for publishing agent-made work, checking review state, reading comments, replying to feedback, and publishing revisions at the same durable URL.
Remote hosted MCP endpoint at https://derive.to/mcp plus a local stdio compatibility package, exposing find/read/catch-up/comment/publish/stage/use/checkpoint workflows to compatible agents.
Market Position
Derive positions itself as an open, Fair Source and self-hostable alternative to model-specific artifact surfaces and chat-based workflows: unlike artifacts/canvases that remain inside an individual AI product, it keeps work, versions, comments, and access controls in a model-agnostic library. It also competes with conventional document/project collaboration tools, differentiating through agent-native CLI/MCP publishing, durable rendered URLs, review-aware revisions, and one-container self-hosting.
Founding Story
Rob Moore said Derive was built by four people after the team saw that AI had dramatically increased output but left less work retained. They identified three problems: agent output was stranded in chats and Slack with no durable home or reliable current version; each new session had to be retaught brand, format, and audience context; and work across many agent tabs felt scattered and difficult to understand. Derive was started to give agent-produced work a permanent home and make the workflow more seamless and fulfilling.
Business Model
Revenue Model
Hosted subscription priced per human editor; viewers and commenters are free and agents do not consume seats. Self-hosting is free under the Fair Source license, while Enterprise offers custom annual contracts and dedicated/residency options.
Pricing Tiers
Up to 3 editors per workspace, unlimited viewers/commenters, CLI/API/MCP, permanent URLs and version history, and 1 GB deduplicated storage.
Unlimited editors, custom domain, white-label shared pages, password links, Brandprint, 50 GB pooled storage, and full analytics history.
Team features plus 250 GB pooled storage, OIDC SSO, audit log, multiple custom domains, guest editor management, uptime SLA, and priority support.
Business features plus a dedicated single-tenant instance, regional/cloud data residency, migration/onboarding, and support contracts.
Target Markets
- Individuals and independent projects
- Small teams
- Teams whose agents ship work requiring review
- Organizations with control, accountability, SSO, audit, and residency requirements
- Developers and teams using Claude Code, Codex, Cursor, or other MCP-compatible agents
- Publishing and retaining AI-generated plans, reports, research briefs, designs, decks, and web pages
- Team review of agent output with accountable inline feedback and approvals
- Cross-model and cross-agent collaboration where context and history must travel with the artifact
- Sales and customer-success workflows using reusable contexts and repeatable branded outputs
- Private internal knowledge/work artifact libraries
- Organizations needing self-hosting, data residency, SSO, audit logs, and dedicated instances