The first three agents to deploy. Each is high-ROI, bounded, and safe to start with HITL.
Agent 1 — Inbox Triage Agent
What it does
Every 2 hours during working hours, scans each employee's inbox and:
- Classifies new emails (Reply-today, Reply-this-week, FYI, Newsletter, Trash)
- Drafts responses for the Reply-today items (using the email-reply-drafter prompt)
- Files newsletters and archives obvious noise
- Posts a summary to Slack DM: "3 replies ready, 8 FYIs, 22 archived"
Never touch flagged / starred / VIP emails without asking
Eval set
5 good + 5 hard test cases per employee's inbox during pilot.
Rollout
Week 1: internal beta with the AI Champion
Week 2: expand to Steering Committee
Week 3: pilot with a full department
Week 4: broader rollout with per-user opt-in
Agent 2 — Meeting-Notes-to-CRM Agent
What it does
After every sales call (detected via calendar + transcription integration):
- Retrieves the transcript from Otter / Fireflies / Gong
- Extracts structured fields: attendees, deal stage, next step, close date, budget signals, competitors mentioned
- Proposes CRM updates for the AE to approve
- Drafts a follow-up email
Autonomy level
Suggest. All CRM updates and emails are drafts requiring AE approval.
20 real recorded calls with known ground truth of what should be extracted.
Rollout
Week 1: 3-AE beta
Week 2: full sales team pilot
Week 3: measure adoption + accuracy
Week 4: refine based on feedback, promote to standard workflow
Agent 3 — Weekly Report Generator
What it does
Every Friday at 3pm, generates a personalized weekly report for each employee:
- Pulls from Calendar, Slack messages sent, tickets/issues closed, docs edited
- Runs the weekly-review prompt (individual/04-prompt-library/19-weekly-review.md)
- Sends the report to the employee's DM
- Optionally rolls up to team dashboards
Autonomy level
Full auto. Report is generated and delivered without human intervention. But it's read-only — no external actions.
Tools required
MCP-02 Calendar (read)
Slack (read own messages, DM output)
Ticketing (read closed tickets)
Docs (read edit history)
Guardrails
Report only to the individual, never shared publicly
No PII of others in reports
Employees can opt out anytime
Eval set
5 employees × 3 weeks of manual weekly reports vs generated. Compare.
Rollout
Week 1: opt-in beta with 10 employees
Week 2: iterate on report format based on feedback
Week 3: broad opt-in launch
Common pattern
Notice the shape:
Bounded scope — one job, well-defined
HITL by default — earn autonomy
Read before write — start read-only, add write behind approval
Eval set before ship — 5+5 minimum, ground truth from real data
Staged rollout — beta → pilot → department → org
Kill switch — every agent can be paused per-user or globally
Anti-patterns
Don't build first:
- Anything that sends external communications autonomously
- Anything that modifies shared systems (CRM records, financials) without HITL
- Anything with a false-positive cost the organization can't tolerate
- "Agent that does everything" — narrow beats broad
Progression: what to build next
After the first three are stable (usually month 3–4):