Skip to Content
🚀 Design Partner Program — Limited spots available →
PlatformConfigureIntelligenceIntelligence

Intelligence

Three screens, one job: tell DigitalCore what to notice, who notices it, and what to do about it.

ScreenQuestion it answers
ThresholdsWhat counts as a signal? When is margin pressure, capacity overrun, or SLA risk worth flagging?
AI AgentsWho is watching? Which assistants are running, what they look at, how often they run, and how proactive they are.
PoliciesWhat can they do without asking? For each assistant and each workflow: allow, block, ask first, or limit.

Tune them as a unit

These three screens are designed to be tuned together. Loosening Policies before calibrating Thresholds creates noise. Adding a powerful agent without setting Policies leaves you blind to what it can do.

1. Thresholds   ->  Get the signal-to-noise ratio right first.
2. AI Agents    ->  Hire the assistants you want and set their cadence.
3. Policies     ->  Decide what they can do automatically vs ask first.

If you tune in the wrong order (granting full action rights before calibrating thresholds), assistants will act confidently on noise.

A concrete example

You want Margin Watch to auto-resolve low-confidence margin signals but ask before opening a decision case:

  1. Thresholds. Set the margin warning level so genuine pressure trips at the right point.
  2. AI Agents. Enable Margin Watch and set cadence to daily.
  3. Policies. For auto_resolve_signal set Allow; for open_decision_case set Asks first.

The result: clean signals, the right specialist watching, and the right level of agency.

Where do I go?

If you want toGo to
Reduce signal noise.Thresholds
Pause all automation now.Policies
See what an assistant did and why.Agent Profiles
Review what a specialist learned.Specialist Memory
Send output to Slack, Teams, or email.Delivery Channels
LibraryAI Agents