The document at the centre
PRD v1.0 — AI-Powered Employee Onboarding Assistant
Readiness 62% · Sections 25 / 30 selected by you
✓13 complete⚠3 assumption to validate🔴9 open question–5 excluded by the pm
Context
- ✓Product Overview
- 🔴Problem Statement
- ⚠Evidence & User Research
- ✓Target Users / Personas
- 🔴Current Experience / Existing Solution
Strategy
- ✓Product Vision & Goals
- ✓Non-Goals / Out of Scope
- 🔴Success Metrics
- ✓Feature Prioritization
Discovery
- ✓Research Insights
- ⚠Assumptions
- 🔴Open Questions / Decisions Needed
Solution
- 🔴Proposed Solution
- ✓User Journeys / User Flows
- ✓UX / Design Requirements
- –Wireframe Requirements
Requirements
- ✓Functional Requirements
- ✓User Stories
- ✓Acceptance Criteria
- ✓Edge Cases
AI
- 🔴AI / ML Requirements
Execution
- 🔴Engineering Requirements
- ✓Analytics & Instrumentation
- ⚠Risks & Mitigations
- 🔴Dependencies
Launch
- –Release / Rollout Plan
- –GTM / Launch Considerations
- –Post-Launch Measurement
- –Product Decision Log
- 🔴Product Readiness Summary
Sections are chosen by you on Customize Your PRD. Excluded sections are never silently regenerated elsewhere.
Problem
GoodNew employees spend their first weeks hunting for basic answers — leave policy, payroll dates, IT access, who to ask — across an intranet, PDFs, Slack and HR email. HR answers the same questions repeatedly.
The cost is not curiosity; it is slow ramp-up and HR capacity spent on repetitive Q&A instead of onboarding design.
AI recommendations
0 openNo open recommendations for this section.
Product decision log
4 decisionsEvidenceInsightDecisionRequirementArtifactOutcome
| Decision | Evidence | Rationale | Outcome measure | Status |
|---|---|---|---|---|
New employees in their first 30 days are the primary user | 5 employee interviews + HR pilot ticket data (62% of tickets from 12 topics) | Highest frequency of information gaps and the clearest measurable cost | Tracked by: repetitive HR tickets per month | Validated |
The assistant answers only from HR-approved sources | HR interviews + Legal risk assessment | A confidently wrong policy answer destroys trust faster than a missing answer | Tracked by: % answers with a cited approved source | Validated |
HR escalation ships in the MVP, not a later phase | Unanswered-question risk observed in the pilot | Preserves trust while knowledge coverage is still incomplete | Tracked by: escalations resolved within one working day | Accepted |
Confidence threshold is tuned by HR, not engineering | None yet — no calibration data | Content owners feel the cost of wrong answers, but tuning needs live data first | Needs: confidence vs. helpfulness calibration data | Open |