Brief Agents

Meet Brief's agents — background workers that watch your connected tools, deliver scheduled briefings, and nudge you when something needs a decision.

Last updated: July 5, 2026

Chat is how you ask Brief questions. Agents are how Brief works for you when you're not asking — they run on schedules in the background, read the data from your connected tools, and deliver findings so your team never ships blind.

Chat vs. agents

When you chat with Brief, you start the conversation. Agents flip that around: each one has a job, a schedule, and a place it delivers results. While you work, they keep your product context current — decisions, signals, competitors, pipeline, roadmap — and surface what matters without being asked.

Every agent is grounded in your connected tools. The more you connect, the more your agents can see.

The Agents page

Open Agents in Brief to see your workspace's roster. Each agent appears as a card showing its status, when it last ran, and what it produced — pending suggestions awaiting your review, items synced, or signals extracted. Click any card to open the agent's detail view.

The detail view has four tabs:

  • Skill — the agent's job description, written in plain language. This is what the agent actually follows on every run, and it's editable: change what the agent focuses on, the tone of its output, or what it should ignore.
  • Options — configuration: the schedule it runs on, which model it thinks with (where the agent supports a choice of the latest Claude models), and the sources it reads from. You can see each source's connection health and jump straight to connecting an integration the agent would benefit from.
  • Grading — the quality rubric the agent's output is held to. You can customize the rubric for your workspace, and reset back to the default at any time.
  • Runs — the full run history: when each run started, how long it took, what triggered it, and whether it produced results, found nothing new, or hit an error.
Editing an agent's skill or rubric is available to workspace admins. Everyone on the team can view agents, their runs, and their output.

How agents deliver their output

Agents write results in one of two ways, and the detail view shows which applies:

  • Suggested for your review — agents working on subjective, strategic material (strategy, positioning, personas, features, decisions) propose changes as pending suggestions. Nothing lands in your product context until someone on your team accepts it. The agent proposes; you decide.
  • Written directly — agents syncing observed facts (revenue, pipeline, work items, competitor pages, signals) write straight to your workspace. A subscription exists or it doesn't — there's nothing to approve.

For review-gated agents, the agent card shows how many suggestions are waiting and links to the review surface.

Reading agent status

Each card carries a status so you can tell health at a glance:

  • Active — the latest run completed recently and produced output
  • Empty — the latest run completed but found nothing new to report
  • Failing — the latest run hit an error
  • Stalled — the agent hasn't produced a decisive result recently
  • Never run — the agent hasn't executed yet

Brief distinguishes real agent failures from infrastructure interruptions (like a routine deploy): an interrupted run shows as an interruption, not a failure, and doesn't mark a healthy agent as broken.

Edits are quality-checked before they go live

When you save a change to an agent's skill, Brief reviews the edit against the agent's grading rubric before it takes effect. Edits that pass become the new live version; edits that would degrade the agent's output are blocked with specific, actionable reasons so you can adjust and resave. The running agent is never affected by an in-progress draft — only an approved version changes what it does on its next run.

Every approved edit is versioned. The Skill tab keeps a full history, and admins can restore any prior version.

Create your own agents

You can create custom agents without leaving chat. Describe what you want watched and where results should go — for example:

"Every Monday morning, digest what our competitors changed on pricing and post it to #competitive-intel."

Brief drafts the agent — its skill, schedule, and delivery target — and shows you an approval card before anything is created. Once you approve, the agent joins your roster on the Agents page like any other, with the same Skill, Options, Grading, and Runs tabs.

Start from a preset

Brief ships a catalog of vetted agent archetypes so you don't have to design from scratch. Ask Brief in chat what agents it can create, or pick a preset directly. The catalog includes:

  • Target account monitor — watch a named account and digest new signals, news, and activity
  • Competitor pricing digest — track a competitor's pricing changes and post them to a Slack channel
  • Keyword mention digest — track mentions of a keyword or brand across your research signals
  • Competitive landscape monitor — digest movement across your whole competitor set
  • Decisions digest — summarize the decisions your team recorded
  • Positioning & messaging brief — digest your canonical positioning into a shareable brief
  • Persona insights digest — surface insights across your persona set
  • Feature roadmap signal — digest roadmap movement across your feature set
  • Pipeline deal digest — summarize deal movement across your pipeline
  • Revenue metrics summary — recap your revenue metrics
  • Research theme digest — cluster recent research signals into themes
  • Strategy brief — tie recent decisions back to your stated strategy
  • Velocity & throughput report — report delivery velocity across your work pipeline
  • Product usage insights — digest usage insights from your product analytics

Most presets take a focus variable — a specific account, competitor, keyword, or area — so the same archetype can run several times with different scopes.

Schedules and delivery

Custom agents run on the cadence you choose — hourly, daily, or weekly, with a time and timezone you set for daily and weekly runs. Results are delivered in-app, and you can route them to a Slack channel: Brief shows you a channel picker from your connected Slack workspace when you set the agent up. See the Slack integration for connecting Slack.

Edit agents from chat

You don't have to open the Agents page to tune an agent. Ask Brief in chat to change an agent — "make the competitor digest shorter" or "move my pipeline digest to Fridays" — and Brief loads the agent, proposes the specific change as a plan, and applies it only after you approve. The same quality check runs on chat edits as on edits made from the Agents page.

Built-in briefings

Two briefings come with every workspace — no setup needed beyond connecting your tools.

Weekly Brief

The Weekly Brief is a strategic digest delivered once a week to every workspace with data: what shipped, what changed, what your signals are saying, and where things are drifting from plan. It's on by default and built as a full tool-calling agent, so it reads your live data rather than a stale snapshot. It appears on the Agents page like any other agent, so you can read its skill, check its runs, and tune its rubric.

Daily Pulse

The Daily Pulse is an opt-in daily summary of activity across your workspace, delivered by email and — if you've connected Slack — posted to your team's notification channel. Each pulse includes an AI-written summary up top and a Discuss with Brief link that opens chat pre-loaded with the digest's context, so you can go from "what happened" to "what should we do about it" in one click.

Turn the Daily Pulse on from your notification settings. The Weekly Brief is on by default; both can be toggled per user.

Proactive nudges

Beyond scheduled briefings, Brief speaks up on its own when something specific needs you:

  • Review invitations — when Brief has accumulated things worth your attention, it opens your next chat session with a short "I noticed a few things since we last talked — want to run through them?" If you've been away from chat for a while and something important is queued, Brief sends a single Slack DM with a link to review together.
  • Decision-expiry reminders — when a decision you recorded is coming up for re-confirmation, Brief emails a short recap: what you decided, why, and what's changed since — so you can re-confirm or revisit with current context.
  • Signal-surge alerts — when a theme suddenly spikes across your research signals, Brief flags the surge so you see the pattern while it's still forming.

Nudges are deliberately gentle. Brief rate-limits how often it interrupts, and deferring a review invitation ("later") quiets nudges for a week while your queue keeps accumulating — you control the cadence, and nothing is lost.

What's next