Brev

The 50+ native integrations are what Brev calls simple one click native connectors, and a logo wall shows roughly forty counterparties. Named systems cover CRM (Salesforce, HubSpot), customer success (Gainsight), engineering tools (Jira, Linear, GitHub), work management (Asana), data and BI (Snowflake, Tableau), support (Zendesk), file storage (Box) and chat (Slack, Microsoft Teams). In the MCP connector panel, Brev, Slack, GitHub and Linear show as connected sources, with Add connector and Manage connectors controls.

The aim is to "bring revenue, pipeline, product, and company goals together and keep them current, automatically." Pulling live progress from CRM and pipeline, engineering trackers, warehouses and support tools into one goal graph is the product's core mechanic.

Data mostly flows in. Brev pulls progress signals from systems of record and writes into its own goal graph, not into Salesforce or Jira. The write side is the delivery channels. In Slack, Brev pitches itself as "like having a 10x operator as your colleague," and a user can manage execution in Brev without ever leaving Slack, a conversational integration that runs both ways.

The integration count appears as 50+ in some places and 30+ in others, twice each.

Sourcebrev.io integrations section and how-it-works step 2, brev.io/mcp connector panelread 2026-09-13

Four agents built around roles (Goal, Meeting, Review and Digest), each with its own product page, hand work between them in a loop of several steps. They collect context from meetings and tools, update the goal graph and detect drift, then draft the review artifact and close the actions. Brev sums up the loop as "reviews drafted, OKRs updated, actions closed, proactively." A running agent panel shows the steps executing in sequence, among them "Updating your company goals," "Tracking dependencies" and "Identifying execution drift".

The sequence is Brev's own fixed operating cadence, and customers cannot compose their own. There is no canvas, graph or step editor, no branching or conditional logic, no versioning or publish lifecycle, and nothing a customer authors, saves or reuses. Customers set their goals, connect their tools and work as usual, and Brev handles the tracking, review prep and follow up. The customer sets the cadence and Brev executes within it, so the cadence decides when the sequence runs, not what it does.

Sourcebrev.io how-it-works steps 1 to 5 and running-agent panel, brev.io/pricing FAQread 2026-09-13

Company Brain is a named knowledge store with its own page, holding "decisions, commitments, and context, always current, always connecting the dots across the company." Teams ask it questions, get updates and see progress across every team, goal, decision or blocker.

The store draws on three sources. A company uploads a strategy doc, OKR sheet or annual plan in any format, and Brev structures the goals, metrics and initiatives from it. The 50+ native integrations pull in revenue, pipeline and product signals as Brev collects context across meetings and tools. Agents also join meetings and turn what is said into structured progress, so what gets said updates the goals.

Updates are grounded in quotes from meeting participants, so a check in traces back to who said it. Synthesized output carries its reasoning inline, for example "Q4 ARR is tracking $1.4M short of pace. Three consecutive weeks of slowing bookings, driven by the CAC spike on paid social pulling new-player volume down." Each such line gives the conclusion, the trend behind it and the driver.

Brev names no index, embedding, chunking or ranking method. Company Brain holds what the company decided, not what the agent did.

Sourcebrev.io Company Brain section, how-it-works steps 3 and 5read 2026-09-13

Some output waits for a person and some does not. Drafted artifacts come to a person first. A notification reads "Your daily brief is ready for your review" with a View full brief control, and Brev talks of reviews drafted, meetings that write their own updates ready before you walk in, and narrative updates drafted as metered work. The person opens the draft, acts on it and keeps the decision on what to do about a flagged risk.

Other work goes ahead without approval. Reviews are drafted, but OKRs are updated and actions closed proactively, and Brev bills automated updates and check ins and automated action items as metered monthly work. Brev names no configurable approval gate and no way for a customer to choose which outputs wait for sign off. It names no approval inbox or queue, no autonomy setting or risk tiering per action, and no escalation path or record of who approved what.

Sourcebrev.io daily brief review notification and how-it-works step 4, brev.io/pricing metered work listread 2026-09-13

The platform is SOC 2 Type 2 certified, and Brev says it is GDPR compliant, CCPA compliant and NIST aligned. It claims ISO 27001 compliance, not certification.

The Enterprise tier adds SSO, SCIM, RBAC and MFA, and SCIM provisioning also appears among the trust claims. These access controls come only with Enterprise, so Growth customers do not get them. Brev promises that "your company's context stays yours," and it does not store raw audio.

Sourcebrev.io trust section, brev.io/pricing Enterprise planread 2026-09-13

Customers can see what an agent is doing now and what it costs, but not what it did before.

Team dashboards, exec ready summaries, cadence effectiveness scores, live progress views and risk signals report on the customer's company, not on the agents, and quotes in updates name who said them.

A Running agent panel names the actor, Brev Operator, and lists its current steps in order ("Updating your company goals," "Tracking dependencies," "Identifying execution drift"). A real time dashboard breaks credit spend down by goal, agent or team, with monthly caps and alerts at any usage threshold, and Enterprise adds organization wide budgets and per team limits. Spend per agent shows how much work each agent did, and the caps control it.

Brev names no run history, task list or execution detail view, no reasoning trace, no error or failure view and no audit log of agent actions. A customer cannot open what an agent did last Tuesday or why it reached its conclusion, and the credit dashboard measures volume, not behavior.

Sourcebrev.io hero running-agent panel, brev.io/pricing spend control FAQread 2026-09-13

Company Brain is the store Brev ships, a queryable knowledge base about the company with its own product page. Brev says "Your company finally remembers everything," and calls the store a single source of truth for execution. Company Brain remembers what the business decided, committed to and blocked on, and teams ask it questions, get updates and see progress across every team, goal, decision or blocker.

Brev names no memory of the agents' own work, such as what an agent did last run, what it learned about the task or what state it carries forward. It names no memory scope per agent or user, no lifetime or purge path and no way to read or write accumulated agent state. Brev says recommendations and coaching grow more relevant over time, but it does not say an agent recalls its own prior run.

Sourcebrev.io Company Brain section and how-it-works step 5read 2026-09-13

Sign up is a hosted app at app.brev.io, the only way Brev is delivered. Brev names no region, cloud provider or tenancy model, claims no data residency, and offers no self hosted, on premises or customer cloud path or any way to choose one.

The Enterprise tier lists bulk credits, customized onboarding, evals and configuration support, forward deployed engineering, and SSO, SCIM, RBAC and MFA, with no hosting choice among them. Organizations with specific security requirements are invited to scope an Enterprise plan with sales.

Brev's promise that your company's context stays yours is about data ownership and training, not about where data sits. GDPR and CCPA compliance and NIST alignment do not fix where the data is located.

Sourcebrev.io trust section, brev.io/pricing Enterprise planread 2026-09-13

Four agents make up the Product menu, each at its own address. Goal Agents live at /okr-agents, Meeting Agents at /meeting-agents, Review Agents at /review-agents and Digest Agents at /digest-agents, with Brev MCP beside them in the menu on every page.

Each agent has its own job, triggers and outputs, and none is a tier or module of another. Digest Agents draft exec summaries on a schedule, while Meeting Agents join a call and turn what is said into goal updates. Dedicated agents for specific workflows are included on every plan, and the metered work table prices their output separately (goals tracked and updated, meetings and reviews run, teams kept live), so each agent's work is a billed unit. A named actor, Brev Operator, appears in the running agent panel.

The Growth and Enterprise tiers both get all four agents. Brev also offers OKR frameworks and review templates, which are document scaffolding, not agents.

Sourcebrev.io product navigation and footer, brev.io/pricing always-included list and metered work tableread 2026-09-13

The agents run unattended, and work starts in four ways. Cadence schedules run automated WBR and MBR cycles, which Brev calls "your business rhythms and rituals, automated," and automated meeting cadences and agenda preparation are billed as metered monthly work. Meeting events start work too. Agents join the meeting, and what gets said updates the goals, so a calendar event starts work that changes system state.

Data conditions surface risks early, with risk and insight signals and reminders and alerts as recurring outputs, and in one example work fires on a trend of three consecutive weeks of slowing bookings instead of on a schedule. Digest schedules prepare daily digests, and a notification says the daily brief is ready for review.

A Running agent panel shows Brev Operator working unattended through "Updating your company goals," "Tracking dependencies" and "Identifying execution drift". Brev takes action and closes the loop, with reviews drafted, OKRs updated and actions closed proactively, and Brev says it keeps the company on track without chasing anyone.

Brev delivers through Slack, Microsoft Teams, any MCP compatible AI client and a CLI, so teams can track goals in Slack, in Teams and in LLMs. Goal based automations can also be started from the CLI or an MCP client.

Sourcebrev.io hero running-agent panel, Meetings & Cadence and Signals & Guidance sections, brev.io/pricing metered cadence itemsread 2026-09-13

The Goal, Meeting, Review and Digest agents run on no named model, provider or model family. There is no model selector or admin control, no disclosed routing and no way to bring your own model or key on either tier.

All the model names Brev uses belong to the customer's side. Brev is completely headless, so teams manage execution inside their favorite AI tool, track goals in LLMs and run goals from Claude, and the goal platform is completely controllable from the customer's AI tool of choice and CLI. Claude, ChatGPT, Cursor and Codex are each the customer's assistant calling Brev through the MCP server. The customer chooses which AI tool drives Brev, not which model Brev reasons with.

Brev describes its own intelligence as AI native, with AI driven summaries, embedded AI coaching and an actor called Brev Operator, and names no provider behind any of them.

Sourcebrev.io Track Goals in LLMs and Run Goals from Claude sections, brev.io/mcp, brev.io/pricingread 2026-09-13

An MCP server and a CLI ship, and MCP access is in early access. The Brev MCP server, a product in the main navigation, is a Model Context Protocol server that gives AI agents like Claude, Cursor and ChatGPT live access to the company's goals, OKRs and operating context. It reads and writes.

Users can read, update and act on company goals from the command line, and typical workflows include checking goal status, posting check ins, querying owners and cadences, and triggering goal based automations, so outside clients can command the platform as well as read it. A CLI ships alongside it, and Full MCP & CLI support is included on every plan.

Every call to action for MCP is Request MCP access or Get access now, both leading to the early access sign up at brev.io/mcp/mcp-early-access. Brev publishes no REST or GraphQL API, SDK, OpenAPI spec or developer documentation site for its own platform. The OpenAI Agents SDK and Anthropic SDK named alongside the MCP server belong to the client, not to Brev.

Sourcebrev.io/mcp FAQ and early-access CTAs, brev.io/pricing always-included listread 2026-09-13

Changes to the agents have no named test step. Brev names no harness or test environment before deployment, no replay of past cases or regression set, no scored outputs, judge, rubric, accuracy or quality metric, and no A/B or shadow comparison of agent versions. AI recommendations, next best actions, coaching and trend analytics are features of the product, not measures of it.

Cadence effectiveness scores rate the customer's meeting rhythm, not the agents' output, and cadence operating system analytics measure the business operating system. Team dashboards, risk signals and progress reports also report on the company.

The Enterprise plan includes customized onboarding, evals and configuration support beside forward deployed engineering for tailored solutions. Those evals are run by Brev's staff during an engagement on a custom priced tier, not shipped in the platform. The $100 in free credits to start is a commercial trial.

Sourcebrev.io/pricing Enterprise plan and metered work lists, brev.io Signals & Guidance sectionread 2026-09-13

Every route into another system is programmatic, through the 50+ one click native connectors, the MCP server and the CLI. Brev calls itself completely headless. It has no interface of its own to drive and does not drive anyone else's, and instead it exposes state and receives commands. There is no browser session the agent drives and no desktop, remote or local computer control.

Agents join standups and business reviews as conferencing participants that take in audio, and the product does not retain raw audio. The CLI is a text interface to Brev's own platform. In Slack and Teams, Brev is a bot in a messaging app that lets users manage execution without ever leaving Slack. Running goals from Claude goes the other way, with a person or the customer's assistant reaching Brev.

Brev's premise is that there is nothing new to adopt because Brev lives in your tools. It reaches systems through their APIs and people through their chat clients, so there is no screen for it to drive.

Sourcebrev.io headless positioning and integrations section, brev.io/mcpread 2026-09-13