A live walkthrough of how I'd structure, govern, and measure Avalara's People & Culture AI portfolio in the first 90 days. Click through the tabs to see each layer.
Business value, risk, feasibility, employee experience. Sequenced into Now / Next / Later / Parked phases. The point isn't to do more; it's to do the right things in the right order, with engineering capacity respected.
AI-driven promotion recommendations are a textbook example of high-risk, low-feasibility work. Bias risk on employment decisions is severe, regulatory exposure is real, and the data infrastructure isn't there yet. Parking it now isn't avoiding the work; it's sequencing it past higher-leverage wins that build trust in the agent first.
Governance should accelerate AI work, not slow it. The right cadence makes risk visible early, keeps stakeholders aligned, and prevents the kind of late-stage surprises that erode trust. Built as a partnership with the central AI/Agentic COE, Legal, Security, and PeopleTech.
P&C AI program team. Status on active initiatives, blockers, decisions needed.
Legal + Security + PeopleTech. New use cases, data access patterns, content review.
P&C leadership + AI/Agentic COE. Sequencing, sunset decisions, capacity reconciliation.
Full P&C function. Outcomes review, strategic re-prioritization, investment asks.
1. What's the human-in-the-loop pattern? No employment decisions without human review.
2. What data does it touch, and at what access level? Read-only by default; write access requires explicit governance approval.
3. What happens when it gets it wrong? Defined fallback paths, escalation logic, and content review cycles.
4. What's the sunset trigger? If it's not delivering measured value within 90 days, what's the off-ramp?
Activity metrics (number of pilots, agents deployed) are easy to game and tell you nothing. Outcome metrics (deflection rate, behavior change, time saved) are what makes the AI story match the success metrics. Tracked weekly, reviewed monthly, presented quarterly.
Weekly internal: Activity + leading indicators for the program team.
Monthly to P&C leadership: Outcome metrics + portfolio decisions made.
Quarterly to executive leadership: Business impact, sequenced wins, investment asks.
Annual: Full portfolio review, sunset decisions, next-year roadmap.
The temptation in a new role is to demonstrate value through volume. The actual job is to demonstrate judgment through sequencing. By day 90, the goal is one shipped initiative as proof, the operating system stood up, and the portfolio roadmap published.
1:1s with Sr. Director People Operations, P&C function leads, PeopleTech/Engineering, Legal/Security, AI/Agentic COE. Map active P&C AI surface area: primary agent capabilities, in-flight work, content libraries.
Inventory existing and proposed use cases. Score on the four-dimension framework. Draft v1 of the portfolio roadmap. Identify the one initiative that's both high-leverage and shippable inside 90 days.
Share v1 portfolio roadmap with P&C leadership. Stand up the governance cadences. Start the bi-weekly Risk & Compliance review. Lock the priority initiative for proof-of-pattern execution.
Execute the priority initiative end-to-end with the team. Run the pilot, measure outcomes against the metrics framework, iterate. Document the operating model so the pattern is repeatable.
Present 12-month portfolio plan to P&C leadership and AI/Agentic COE. Publish results from proof-of-pattern initiative. Sequence next two priority initiatives. Establish quarterly executive readout cadence.
P&C leadership knows what's coming, what's parked, and why. PeopleTech knows what's on their roadmap. The AI/Agentic COE has a clear partner. One shipped initiative has built credibility. The metrics framework makes future decisions easier. And the operating model means this work continues even when I'm not in the room.