What you’ll do & how you’ll make your mark.
The agent experience architecture. Define the end-to-end model for how agents are discovered, invoked, handed off between, and held accountable to the customer. Own the shared primitives — identity and permissions, context and memory, intent routing, hand-off and escalation, human-in-the-loop approval, transparency and traceability, guardrails and fallback.
Connecting the dots to the customer. Translate a portfolio of internally-owned agents into a single customer narrative: what the customer can ask for, what happens when they do, how they know it worked, and what it costs. Own the customer journey across agents, not just the agents.
The interface surface strategy. Set the strategy for how agentic capability is exposed across:
•
Portal — conversational and ambient agent surfaces inside the logged-in experience, alongside and eventually in place of traditional navigation
•
APIs — clean, agent-consumable contracts over our commerce, hosting, domains, and account systems
•
MCP — our capabilities as tools that both our own agents and external agents (Claude, ChatGPT, IDEs, partner platforms) can call safely
•
Connectors & channels — bringing our agents to where the customer already works, and bringing their tools into ours
Standards and the operating model. Establish the shared patterns, quality bars, and interaction standards that let a dozen teams ship agents independently without fragmenting the experience. Define what “good” looks like — latency, accuracy, containment, escalation, tone, disclosure — and the review mechanism that holds it.
Measurement. Define the metric framework for agentic experience: task success and completion rate, hand-off and containment, time-to-value, trust and intervention rates, and the downstream commercial impact on conversion, attach, and retention.
Roadmap and influence. Build the multi-quarter roadmap for the connective layer, and drive alignment across product, engineering, data, design, care, and marketing to sequence it. Partner especially closely with Data & Personalization — agents are only as good as the context and signals they act on.