How-to: CoverGuard Agents (automations)

Your own AI workers that run jobs on a schedule — pre-bid checks, best-coverage scans, renewal sweeps — and send you a plain-language briefing, so routine monitoring happens on its own.

Who this is for: Everyone — every plan includes the out-of-the-box agent fleet. Building your own custom agent is a Pro-tier feature. Where: the Agents tab (/automations).

Key thing to know: they ship OFF (with one exception)

Every out-of-the-box agent is provisioned PAUSED for every account — nothing fires automatically until you activate one. This is deliberate: a new user isn't flooded with background jobs. Your Agents tab also leads with just the recommended few for your role (≈3 for buyers, ≈5 for agents), not the whole catalog — the rest are one click away in the "Add agent" gallery.

The one exception: if you're a new home buyer, your Best-Coverage Optimizer agent starts already active and runs weekly from day one — no setup needed. It's the agent that most directly rewards your first action here (searching and saving a property), so you see a real result without having to know to turn anything on. Home sellers and every other role still see this (and every) agent paused as above.

Activate a recommended agent

1. Open Agents.
2. Pick a recommended agent and click Activate (this is a PAUSED → ACTIVE change).
3. It now runs on its schedule (hourly/daily/weekly/monthly), starting at the next real boundary.

Add another from the gallery

Click Add agent, browse the gallery (filtered to your persona for demo operators, full catalog otherwise), and one-click install. Gallery installs start active.

Create a custom agent

Click New agent, give it an objective (what it should do) and a schedule. Custom-agent creation is a Pro-tier feature. Every agent — out-of-the-box and custom — is bound to CoverGuard's standard governance contract (role, scope, output shape, escalation), visible as a collapsible panel on the agent's detail view.

Agent types

When you build an agent, you start by picking its type, and the type's name becomes the agent's starting name (change it to whatever you like): Property Monitor (watches a property and reports changes), Carrier Appetite Tracker (tracks carrier activity in the markets you watch), Carrier Tracker (flags carriers that stop writing), Lead Qualifier, Risk Analyzer, Portfolio Summarizer, or Custom Agent.

These are agents you build, run and pause. They are separate from the three always-on agents in the Always on section at the top of the tab (Market Watch, Renewal Guard and Savings Finder), and no type or gallery agent shares a name with them, so you can always tell which one you are looking at.

Point an agent at specific properties, clients or carriers

Under the objective — when you install from the gallery or build a custom agent — pick what the agent should focus on from your own records: Add a property searches your properties by address, Add a client searches your clients, and Add a carrier searches carrier names. Each pick adds one line to the objective, such as Scope: the carrier Acme Mutual., and the objective stays yours to edit. A client or carrier that isn't on file still works: type the name and choose Add client or Add carrier. Agents that watch ZIP codes rather than properties don't show these pickers.

Run one now

Click Run now on any agent to execute it immediately and watch the result live.

Briefings, notifications & email

When a scheduled run finishes, you get:

• An in-app notification (bell) + web push with a plain-language snippet and a deep-link that opens that agent's run history.
• On success, optionally a full-briefing email — but only if both (a) your global "Agent briefing emails" master switch is on (Account → Settings → Notifications) and (b) you've opted that agent in. Email is the noisier channel, so it's off by default; the bell + push always fire.

Briefings are written for a non-technical reader — no jargon, no raw score dumps.

(Manual "Run now" results show live and don't notify/email.)

Manage an agent

Open an agent to rename, re-schedule, pause, view run history, or toggle its briefing email. Delete removes it (a deleted non-recommended agent stays gone; a deleted recommended one is re-seeded so critical workflows are never stranded).

Requirements for scheduled runs to actually fire

Scheduled runs execute on CoverGuard's background workers and need a working AI key — the platform's managed key, or your own if you've connected one under Account → AI & Models. If a scheduled run didn't fire, it's usually a temporary worker/queue hiccup: try Run now, and contact support if it persists.

Troubleshooting

• "My agent didn't run" → most often it's still paused (activate it), or the queue drainer isn't configured. See AI & agents.
• Run failed → check the run's recorded cause; a common one is the AI key being unset, or the job queue tier not provisioned on the connected database.
• No briefing email → check both the master toggle and the per-agent opt-in; FAILED runs are never emailed.
• Can't create a custom agent → it's Pro-tier. See Billing & plans.

Do this next

Open Agents, pick the recommended agent that matches your day, and Activate it — the fleet only works for you once one is live.

• CoverGuard Advisor & BYOK — the assistant that powers your agents
• Alerts: carrier-exit & renewals — the proactive monitors agents complement
• AI & agents troubleshooting — when a run doesn't fire