Pencel

Appearance

Pulse ​

Pulse is your home for reading agent output and acting on it — approvals, follow-ups, drafts, and your personal to-dos, all in one place.

Pulse is where results land. Your agents publish briefing and roll-up cards here; you read them in place and act without hopping between screens. The URL is /operations; the sidebar entry is Pulse. (It absorbed the old split between a "Feed" home and an "Operations" control plane — run history and schedules moved to Jobs.)

Scopes ​

Pulse organizes work into a few scopes:

  • Active — anything happening now or waiting on you: runs in progress and steps paused on approvals or input requests.
  • Feed — recent results as briefing cards and roll-up cards (the default scope). This is where completed activity lands for you to read and act on.
  • To-dos — your personal checklist. Capture a task, check it off, or hand one to an agent to do now (a one-off run). See To-dos.

When you install a solution pack that produces records (like the RFP pack's opportunities), Pulse adds a pack-record scope — a dedicated tab for that record type (e.g. Opportunities). A record-emitting run collapses into a single roll-up card in Feed that routes you into its scope tab, where per-record actions (like Write proposal → Tailor) live. Non-pack users never see these extra tabs.

Where did Scheduled and History go?

Recurring schedules and full run history now live on the dedicated Jobs surface (/jobs), which owns run/operate concerns — monitoring, trust rollups, and scheduling. Pulse is for reading and acting on results; Jobs is for watching things run.

Briefing and roll-up cards ​

The Feed reads like an email inbox: a compact list of cards on the left (agent avatar, topic, headline, teaser, and a few fact chips) and a wider reading pane on the right that shows the full briefing for the card you select — highlights, roll-up previews, the recommended next step, source links, and why you're seeing it. Silent runs (nothing changed) suppress their card but still record a run so the history stays complete. You can sort and filter the list; importance is one ordering option, not the only one.

Each card offers inline actions without leaving Pulse:

  • Ask — ask the agent a follow-up about the result (a quick read-only sub-run).
  • Reply — draft an editable reply to the card's source (e.g. Reply in Gmail / Slack). Because replies send externally, they're held behind a review gate so you approve the exact content before it goes out.

Radar insight cards also appear here — see Radar.

Add Task ​

The Add Task button gives you two ways to put an agent to work:

  • Build it into an automated task — hand your goal to the guided Your job flow. Pencel compiles it, tests it safely, and schedules it as a reliable recurring job (landing on the Jobs page). Use this for anything you want to run on a cadence — inbox triage, news scans, morning briefings.
  • Run once — an ephemeral one-off: describe what you want and an agent does it now, no workflow saved. Use this for a quick "ask my agent" without committing to a recurring task.

Recurring tasks run on a schedule (cron under the hood; see Triggers & Schedules for Heartbeat intervals). When a scheduled run finds nothing new it can stay silent — the run is still recorded, but no card is published — so a fast cadence doesn't flood the Feed.

Run history and schedules ​

Run history, live monitoring, and schedules live on the Jobs surface, not in Pulse. Recent runs are kept in the local database with their full event log; cost, latency, and verifier verdict are surfaced on each row, and you can open the inspector for the full trace or retry from a failed step.

  • Jobs — run history, monitoring, trust rollups, and schedules
  • To-dos — the personal checklist scope
  • Radar — opt-in background scans that post insight cards
  • Triggers & Schedules — Heartbeat and cron cadences behind recurring tasks