Job Description

Strategy and portfolio

– The group’s product vision, direction and strategy for a 12–18 month roadmap: what gets built, in what order, and why.

– Portfolio prioritization and capacity allocation, including how many tools the group can responsibly carry at once.

– The build-versus-buy recommendation for every initiative, against a buy-or-configure-first default, and the business case behind it.

– The sunset decision for tools that do not earn their adoption.

– A two-to-three year view of how AI changes product and engineering work, and a roadmap that stays consistent with it.

Building AI products

– End-to-end product definition for the group’s AI tooling: problem, users, the workflow it replaces, the adoption path, the measurement plan, the maintenance owner.

– AI-native specifications a strong engineer can build from — task boundaries, context sources, failure modes, guardrails, human-in-the-loop points, and the quality bar in numbers.

– The product calls that shape the architecture: workflow versus agent, retrieval versus fine-tuning, model selection per use case, and the cost and latency budget.

– The quality bar and evaluation strategy: what "good enough" means before the build starts, a failure taxonomy built from real usage, and the regression discipline when prompts or models change.

– How the human enters the loop, how uncertainty is shown, and what happens when the system does not know — the decisions that determine whether people trust the tool.

– The incident and rollback plan for non-deterministic failure, written before launch.

Adoption

– The adoption outcome, measured as instrumented depth of use — not seats provisioned or enthusiasm in a demo.





– The adoption path for each tool: pilot teams, a champion in each team, onboarding, office hours, handover to support.

– Evangelizing the work across product and engineering: the demo, the prototype, the case made repeatedly and well.

– Facilitating the sessions where practice actually changes and leaving them with commitments rather than sentiment.

– Change management, rollout, training and enablement content.

– Honest reconciliation of instrumented usage against self-reported benefit, and ownership of the gap.

Practice and craft

– The product operating standards for the organization — intake, prioritization inputs, PRD conventions, definition of done, documentation, decision log — kept deliberately light.

– Coaching and mentoring product managers, particularly those earlier in their careers, in continuous discovery, outcome framing, evidence-based decision-making and AI-native practice.

– The AI literacy curriculum for the product organization — designed and taught or outsourced.

– Recurring forums where teams share what they discovered and what they decided.

Measurement and reporting

– The definition, baseline and instrumentation for the programme’s success metrics, quantitative and qualitative.

– A metric-led reporting cadence to senior leadership, including the results that did not work.

– The trade-off framework — speed, reliability, cost, data risk — communicated in writing.

Governance partnership

– The product-side data decisions: what data each surface accepts, which model serves which use case, and the guardrails that go with it.

– Partnership with Legal, Privacy, Security and IT on data classification, model approval, access and responsible-AI standards.

📌 AI Product Ops & AI Enablement Lead (Barcelona)
🏢 NielsenIQ
📍 Barcelona