Agent skill
Automation Wedge Scan.
This free skill helps an agent inspect approved internal context and identify the first bounded workflow where automation could save time, improve quality, or create measurable operating value.
It is designed for read-only discovery. The agent should return a review packet for a human, not change systems, message customers, or launch automations.
How to use it
Give the agent read-only access to the easy sources first: Slack, Google Drive, meeting notes, CRM exports, helpdesk tickets, shared spreadsheets, docs, and policies. The agent should look for repeated work with judgment inside it, rather than only high-volume clerical activity.
The scan is strongest when the human names a business area first: sales follow-up, onboarding, renewals, implementation, support, collections, QA, or internal operating cadence.
Expected output
- Candidate wedges
- Three to five repeated workflows where AI could prepare, check, draft, route, monitor, or escalate bounded work.
- Value estimate
- A rough time-saved or value-created range, with assumptions and confidence stated plainly.
- Context map
- The Slack channels, Drive folders, docs, CRM records, tickets, spreadsheets, meetings, or policies the agent needed.
- Risk boundary
- The permissions, review points, audit trail, human approvals, and quality checks required before production use.
- First 30 days
- A short build path for validating one wedge without turning it into a broad transformation program.
Boundary
Do not let the agent send messages, modify records, create API keys, move files, or change permissions during the scan. The output is a decision packet for a human owner.
If the agent cannot establish a process owner, a measurable result, and a safe path to production, the correct answer is that the wedge is not ready yet.