Use cases

The best Doozies own durable areas of work, not isolated prompts. Use these examples as starting points, then adapt the role, connections, evidence requirements, and approval boundaries to your organization.

Chief of staff team

Owns: cross-functional follow-through, decision tracking, meeting preparation, and a source-linked digest of what needs attention.

Connect: calendar, email, Slack, meeting captures, planning documents, and to-do systems.

Start with:

Review this week’s meetings, commitments, and active projects. Produce a Friday brief with decisions made, overdue follow-ups, risks, and the five things that need my attention. Create internal to-dos, but draft external follow-ups for approval.

Use multiple Doozies when the scope spans functions: an Executive Operations Doozy can coordinate while project-specific Doozies investigate their own areas in parallel.

Sales outbound

Owns: account research, contact prioritization, personalized drafts, and follow-up preparation.

Connect: CRM, product-intent data, approved research sources, email, and calendar.

Start with:

Find the ten accounts with the strongest new buying signals. Explain the evidence, identify likely stakeholders, and draft one useful outreach angle for each. Don’t send messages or change CRM stages without approval.

Require source links and prohibit invented contact details. Separate research, writing, and CRM operations among Doozies when volume grows.

Talent scouting

Owns: candidate research, pipeline organization, outreach drafts, interview preparation, and scheduling coordination.

Connect: applicant tracking system, approved sourcing tools, email, calendar, and role scorecards.

Start with:

Build a review-ready shortlist for the senior platform role. Score candidates against the attached rubric, show evidence for every score, flag uncertainty, and draft outreach for the top five. Don’t contact candidates.

Keep employment decisions and sensitive personal data under human review.

Paid media

Owns: campaign monitoring, anomaly investigation, reporting, and budget recommendations.

Connect: advertising platforms, analytics, attribution data, budget spreadsheets, and Slack.

Start with:

Compare the last seven days with the previous four-week baseline. Investigate material changes in spend, conversion rate, and acquisition cost. Draft budget moves with expected impact, but don’t change campaigns.

A schedule (aka routine, recurring task, or timed automation) can run the analysis daily while an approval rule protects budget and publishing changes.

Expense management

Owns: reconciliation, missing-receipt follow-up, coding suggestions, and review preparation.

Connect: expense platform, email, shared drive, accounting system, and finance spreadsheets.

Start with:

Reconcile this week’s card transactions against receipts and policy. Prepare a table of matched items, likely categories, duplicates, and missing information. Draft follow-ups but don’t submit or approve expenses.

Financial submissions, payments, and account changes should always remain approval-gated.

Product performance

Owns: targeted performance investigations backed by logs, traces, analytics, and reproducible evidence.

Connect: observability, product analytics, incident tooling, source control, and the Morph repository.

Start with:

Investigate the checkout latency increase since Tuesday. Correlate releases, traces, database timing, and client analytics. Return the most likely causes, contradictory evidence, and a proposed fix with a verification plan.

An investigator Doozy can gather evidence while an engineering Doozy evaluates the code and another prepares the incident summary.

Bug reproduction

Owns: turning incomplete reports into reliable reproduction packs.

Connect: issue tracker, staging environment, browser tools, logs, and source control.

Start with:

Reproduce the issue described in ENG-418. Record the exact environment, actions, observed behavior, console and network evidence, and the smallest reliable reproduction. Don’t modify production data.

Use Morph snapshots or clean branches to keep tests repeatable.

Account health

Owns: risk and expansion signals across customer activity, support, billing, and relationship history.

Connect: CRM, product usage, support platform, billing, call notes, and email.

Start with:

Review the enterprise portfolio and identify accounts whose health materially changed this month. Explain each signal, separate facts from inference, and draft a recommended next action for the account owner.

Keep customer-facing messages and contractual changes behind approval.

Turn an example into durable ownership

Once a to-do (aka task or work item) works well:

1. Save reusable instructions and examples as a skill.
2. Add the ongoing outcome as a responsibility.
3. Set working hours and an appropriate schedule or event trigger.
4. Define evidence and review requirements.
5. Keep consequential external actions approval-gated.

The goal is dependable ownership with a clear place for human judgment.

Frequently Asked Questions