Skip to main content
OverviewVirtual EnvironmentAgent InboxWalker Agents

Agent Workflows

Walker Agents

Walker Agents are built-in specialists that watch your data and propose actions. You do not build them on a canvas: they produce suggestions that queue up for you to execute, pause, or reject. Each agent also carries an autonomy level — shown on its card as human guided, semi-autonomous, or autonomous — which governs how much it does before asking.

How they differ from workflow agents

  • You do not build them. A workflow agent is a flow you assemble in the Agent Suite. Walker Agents ship with the platform as specialists — paid ads, SEO, content generation, and audience intelligence.
  • They propose rather than execute. A workflow agent runs and does what it was built to do. A Walker Agent produces a suggestion that sits in a queue until you execute, pause, or reject it.
  • Their output is a recommendation, not an artefact. Each suggestion carries a title, a description, a priority, a confidence score, and an estimated revenue effect — enough to judge it without opening anything else.
  • They share the same surfaces. Walker Agents appear on their own tab on the My Agents screen alongside your custom agents, and their conversations appear in the Agent Inbox with the source marked as Walker.

Working the suggestions queue

The Walker Agent dashboard is a list of suggestions filtered by status — pending, approved, executing, executed, paused, or rejected — with totals above it for how many suggestions are in view, their combined estimated revenue, and their average confidence score. Each suggestion shows which agent type produced it and its priority. For each one you can:

  1. Execute — accept the suggestion and let it proceed.
  2. Pause — hold it without rejecting, if it is right but not now.
  3. Reject — decline it.

Suggestions can also be selected in bulk and actioned together, which is the practical way to clear a backlog of low-priority items. Suggestions arrive in batches, so the queue tends to fill in groups rather than one at a time.

Whether they are earning their place

The Walker Agent analytics screen reports over a time range you select:

  • Total suggestions, and how many are pending, executed, rejected, or paused.
  • Total estimated revenue across all suggestions, and the estimated revenue of the ones you actually executed.
  • Average confidence, and your acceptance rate.
  • Breakdowns by agent type and by priority, plus a timeline of suggestions, executions, and rejections.

Acceptance rate against average confidence is the useful pairing: an agent producing confident suggestions you keep rejecting is mis-calibrated for your brand, and worth narrowing rather than ignoring.

How suggestions reach you

A notification bell in the app surfaces pending suggestions and lets you act on them without leaving the screen you are on. Outside the app, the Walker Agent settings control the channels: email can be enabled with a delivery frequency, WhatsApp can be enabled after you verify your number with a code sent to it, and quiet hours suppress notifications between a start and end time you set.

Where to find them

The Walker Agent screens are not a top-level item in the main sidebar. You reach them from the notification bell when suggestions are waiting, and from the best-practices panel in Content Studio. Walker Agents themselves are visible in the Agent Workflows area: they have their own tab on My Agents, where each one shows its status, health, and capabilities and can be triggered or configured like any other agent, and their conversations appear in the Agent Inbox filtered by source.

Related: Agent Inbox for threads and approvals, and The Virtual Environment for agents you build yourself.