Brief
Brief is a product context layer for teams whose AI coding agents build without strategy. It captures decisions from Linear, Jira, Notion, Slack and GitHub, then serves them back through MCP, chat and a CLI.
What is Brief?
Brief calls itself a Navigator: a layer of product context and recorded decisions sitting between the tools a team already uses and the AI agents that write its code. The problem it addresses is stated bluntly on the homepage — coding agents ship fast but build blind, with no strategy, no memory of what was decided last week and no access to customer research. The product has three parts. A web app acts as the command centre where you view the Product Graph and search decisions. An MCP server and CLI carry that context into Cursor, Claude Code and Windsurf. And integrations with twenty-nine business tools capture decisions where they actually happen. The cycle runs Capture, Connect, Navigate: Brief reads your tools and extracts decisions, those decisions build a searchable graph, then your team and agents query it from wherever they work. What the graph holds is specific: product strategy such as ideal customer profile, competitive advantages, goals and metrics; a decision log recording what was decided, why, and when it expires; customer context including research, feedback themes and personas; the work pipeline; and team velocity. Agents reach it through three MCP tools — brief_get_onboarding_context to load a workspace, brief_search for facts, and brief_ask for judgement, with a check mode for architectural forks. The site publishes its own benchmark figures, drawn from a paper it hosts: 95% decision compliance when agents can reach Brief against 46% with the codebase alone, eight of eight merge-ready tasks against two of eight, and a 68% drop in cost per merge-ready task. A separate page estimates roughly 15% less token spend and around 30% faster time to market. The publisher is Brief Labs Inc., based in San Francisco, with a four-person team led by founder Drew Dillon. The stack is declared as Next.js, TypeScript, PostgreSQL, Anthropic Claude and the Vercel AI SDK. Brief is explicit about what it is not: neither a task tracker, nor a documentation tool, nor a code assistant, nor a general chatbot.
What it does
- Capture product, technical and process decisions automatically from the tools your team already works in
- Serve that context to Cursor, Claude Code, Windsurf, VS Code and any MCP-compatible assistant, straight from the IDE
- Search decisions, rationale and expiry dates by natural language, for humans and agents alike
- Check a proposed architecture, auth, database, API or breaking change against what was already decided
- Build a Product Graph covering strategy, personas, research signals, work pipeline and team velocity
- Answer product questions from chat on the web, in Slack, in Microsoft Teams, on mobile or by voice
- Run agents that deliver a weekly digest and flag decisions due for review or signals that are surging
When to use Brief / When not to
A quick filter to help you decide if Brief is the right fit.
When to use Brief
- Technical founders acting as their own product manager, from pre-seed to Series A, who carry the roadmap in their head and need it written down somewhere their team can query
- Engineering leaders whose developers ship fast with Cursor, Claude Code or Windsurf but whose agents keep proposing approaches the team already rejected
- Product managers who now build with AI and need their assistant to know the decision history, research signals and constraints before it suggests anything
- Teams already invested in Linear, Jira, Notion, Slack and GitHub who want a context layer on top rather than yet another tool to keep up to date
- Go-to-market teams whose customer evidence lives in Gong, Fathom or Fireflies calls and in HubSpot or Attio records, and who want engineering to see it
When not to use Brief
- Anyone looking for a task tracker: Brief states plainly that it is not one and points you to Linear, Jira or Asana instead
- Teams wanting a documentation tool: Brief records why something was decided, not what exists, and refers you to Notion or Confluence for the latter
- Developers shopping for a code assistant: Brief does not write code and names Cursor and Copilot as the tools for that job
- Solo users or duos who need centralised billing and an admin dashboard, since those start at the Teams plan with a five-seat minimum
- Organisations that require a published SOC 2 or ISO 27001 certificate, a signed uptime commitment or an EU establishment before they can sign
How to use Brief
A typical end-to-end flow, from setup to results.
- Create a free account at app.briefhq.ai/sign-up; the Free plan needs no credit card
- Connect the tools your team already lives in, choosing which repositories, channels and spaces Brief may reach
- Let Brief read those sources and assemble your Product Graph, which the site says starts producing value within minutes
- Point an MCP-compatible host at https://app.briefhq.ai/mcp over Streamable HTTP; OAuth opens on the first tool call and there is no API key to create or rotate
- Install the command-line client with npm i -g @briefhq/cli if you prefer working from a terminal
- Follow the host-specific guide for Cursor, Claude Code, VS Code, Claude Desktop, Codex, ChatGPT or Windsurf in the MCP setup documentation
- Have your agent call brief_get_onboarding_context at the start of any product task, then brief_search for facts and brief_ask for trade-offs
- Run brief_ask in check mode before an architecture, auth, database, API, dependency or breaking change
- Record new decisions as they are made, from chat, from Slack with the /decide command, or from your agent, so the next person inherits the reasoning
- Upgrade from Settings then Billing when you need more credits, seats or the admin dashboard; checkout runs through Stripe and a workspace admin must complete a Teams purchase
Pros & Cons
Pros
- Fits into the tools you already use rather than replacing them, so no workflow change is required to start
- Standards-based access through MCP with OAuth and dynamic client registration, meaning no API key to create, copy or rotate
- Reachable from many surfaces at once: IDE, web, Slack, Microsoft Teams, mobile, Chrome extension and voice
- Twenty-nine integrations spanning nine tool families, including CRM, meeting transcripts, analytics and databases
- A permanent free plan with fifty AI credits a month and no credit card, which makes evaluation genuinely risk-free
- Unusually detailed privacy policy: twelve named subprocessors with their locations, retention set out category by category, and thirty days' notice with a right to object before a new subprocessor is added
- Machine-readable documentation published openly, including an OpenAPI specification, an MCP server manifest, an agent index and an Agent Skills descriptor
Cons
- No security certification is published — neither SOC 2 nor ISO 27001 — and the Trust Center advertised in the footer was unreachable when this record was compiled
- No service level agreement and no uptime commitment, stated as such in the terms you accept on sign-up
- Brief trains its own specialised models on customer-isolated interaction data, and stepping out of that means emailing privacy@briefhq.ai rather than flipping a setting
- AI credits are capped and overages are billed at $0.10 per credit on Pro and $0.08 on Teams, so the real monthly cost depends on usage
- Centralised billing and the admin dashboard begin at the Teams plan with a five-seat minimum, and single sign-on only at Enterprise on quotation
- Mobile is beta: an iOS TestFlight build for voice, portrait only, with document editing still desktop-only, and nothing published on the App Store or Google Play
- Corporate transparency is thin — no full postal address beyond a city, no Article 27 representative for a service that explicitly addresses EU and UK users, and a domain only registered in August 2025
Pricing & Plans
A permanent free plan is available at $0 with fifty AI credits per month and no credit card required. The lowest paid entry point is the Pro plan at $49.00 USD per seat per month on annual billing, with the Teams plan at $89.00 USD per seat per month and a five-seat minimum. Enterprise pricing is on quotation.
- 50 AI credits per month
- shared workspace
- pooled context
- MCP integration
- 200 AI credits per month per seat
- $0.10 per credit overage
- per-user billing
- email support
- one seat minimum
- 500 pooled AI credits per month per seat
- $0.08 per credit overage
- centralised billing
- admin dashboard
- priority support
- five seats minimum
- negotiated AI credits
- single sign-on and security controls
- dedicated Customer Success Manager
- SLA-backed support
- multi-workspace capability
Data, GDPR & hosting
A consolidated view of how Brief handles your data.
GDPR overview
Concrete provisions are in place, though the site never claims compliance in so many words. Brief Labs Inc. names itself data controller and designates a data protection officer, Andrew Dillon, reachable at privacy@briefhq.ai. Section 11.4 of the privacy policy enumerates the full set of EU and UK rights: access, rectification, erasure, restriction, portability, objection, withdrawal of consent and complaint to a supervisory authority, with a one-month response window extendable to three. Transfers to the United States rely on the European Commission's Standard Contractual Clauses 2021/914 and the UK International Data Transfer Addendum, backed by transfer impact assessments. A Data Processing Addendum is offered on request. Two gaps stand out: no Article 27 representative is designated in the Union, and no security certification is published.
Who owns the data?
Under the Terms of Service, you keep every ownership right over the content you create or upload, which Brief calls "Your Content". You grant Brief Labs Inc. a worldwide, non-exclusive, royalty-free licence to store, display, reproduce, modify and distribute that content, but strictly to deliver the service and to process it with AI models. The licence survives deletion for what the terms call a commercially reasonable period, so that pending processing and backups can complete. The platform itself, its software, design and trademarks, remains the property of Brief Labs Inc. Internally, only the chief executive holds administrative access to customer data, and only for support requested by the customer.
Reuse rights
Your content stays yours to reuse freely: nothing in the terms restricts what you do with the documents, decisions and product graph you build, and no permission from Brief is needed to export, republish or repurpose them. On Brief's side, the licence is bounded by purpose. The vendor states it does not feed customer content to third-party AI models and uses enterprise API tiers with zero data retention where available, though it does train its own customer-isolated specialised models on interaction data, never shared across customers. Aggregated and de-identified data may be derived from your content. If you would rather no AI processing happened at all, the privacy policy invites you to write to privacy@briefhq.ai to discuss alternative configurations.
Data retention & training
Hosting summary
Brief is operated from the United States and the bulk of the data sits there. Amazon Web Services provides cloud hosting, database, object storage and infrastructure, located in the United States, with EU residency available on request. PostHog handles product analytics from the United States or the EU, with EU residency options. The ten remaining named subprocessors are United States based: Anthropic, OpenAI, ElevenLabs, Clerk, Stripe, Inngest, Sentry, Postmark, Firecrawl and Cloudflare, the last operating a global network. Transfers of European and British personal data rely on the European Commission's Standard Contractual Clauses 2021/914 and the UK International Data Transfer Addendum, with transfer impact assessments where no adequacy decision applies. The public site itself resolves to a United States address on an anycast network. No hosting certification is published, and the Trust Center advertised in the footer was unreachable when this record was compiled.
Where Brief works
Country-level availability.
Not available in
Things to keep in mind
Risks and trade-offs to weigh before adopting Brief.
- Connecting Brief hands it an unusually wide field of view: source code from GitHub and GitLab, tickets, Notion and Confluence pages, Slack messages, meeting transcripts, CRM records and even Supabase or PostgreSQL databases. Scope each integration deliberately; resource-level scoping exists, but only if you use it.
- Brief trains its own specialised models on customer-isolated interaction data. It is not shared between customers, but stepping out requires an email exchange rather than a switch, and the default is participation.
- OpenAI retains API inputs and outputs for up to thirty days for abuse monitoring, and the privacy policy says this may include document extracts, issue descriptions and contact names pulled from your integrations. That window exists whatever Brief's own retention says.
- There is no uptime commitment and no service level agreement, so a team that makes Brief the entry point to its product knowledge is accepting an availability risk with no contractual remedy.
- Administrative access to customer data sits with a single person, the chief executive. That is a small attack surface and a single point of failure at once.
- Delegating the memory of why decisions were made can quietly erode the habit of holding that reasoning yourself. A team that stops arguing about trade-offs because the graph already has an answer has lost something the tool was never meant to replace.
- The subscription renews automatically, credit overages are billed on top, and cancelling refunds nothing for the unused portion of the current period. Disputes fall under Delaware law and AAA arbitration in San Francisco, with a class action waiver you have thirty days to opt out of in writing.
Setup & Integrations
Technical difficulty
Low for the web app, moderate for the agent side. Signing up is self-service on the free plan with no credit card, and the quick start is advertised at five minutes. Connecting an MCP host means pasting one URL, since OAuth opens on the first tool call and there is no API key to manage, and the command-line client installs with a single npm command. The real friction is elsewhere: you need administrator rights on the third-party tools you want to connect, and only a workspace admin can buy a Teams plan.
Deployment
Integrations
Behind Brief
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement Brief.
Frequently asked questions
Is Brief another project management tool?
How is Brief different from documentation?
Which AI assistants does Brief work with?
Is there a free plan, and what does the first paid tier cost?
How many integrations are available?
Is my data used to train AI models?
Where is my data hosted, and can it stay in Europe?
How long is my data kept, and how do I delete it?
Is a Data Processing Addendum available?
Is there a mobile app, and what is the minimum age?
Should you pick Brief?
Brief answers a problem that has become concrete for anyone shipping with coding agents: the assistant writes plausible code against constraints it has never been told about. Its response is a decision layer, where each choice is stored with its reasoning and an expiry date, then exposed over MCP so an agent can consult it at the moment of the fork. That shape is genuinely different from a rules file or a vector search, both of which carry text rather than judgement, and the twenty-nine integrations mean the record fills itself from tools the team already uses rather than from discipline nobody sustains. The caveats are those of a young company. The domain was registered in August 2025, the team numbers four, no security certification is published, the advertised Trust Center did not answer when this record was compiled, and the terms concede there is no uptime commitment. The vendor also trains its own specialised models on customer-isolated interaction data, and opting out means writing an email rather than changing a setting. Corporate transparency stops at a city name, with no full postal address and no Article 27 representative for a service that addresses EU and UK users directly. Against that, the privacy policy is unusually explicit for a company this size, naming twelve subprocessors with their locations and setting out retention category by category. And the permanent free plan removes the usual reason to hesitate: fifty credits a month, no credit card, and a connection that takes an OAuth handshake rather than a key. Try it on one real project before committing budget, and if you operate under European constraints, ask about EU residency and the DPA before you sign.
- Choosing a selection results in a full page refresh.
- Opens in a new window.