Building the AI-Native Enterprise: Groupon's VP on the 'AI Factory' and Working With Claude Co-work

Masha Sharma is VP of merchant experience at Groupon, where she's rebuilding the company into an AI-first growth engine for its merchants. AI-native, in her framing, means agents do the actual work while humans direct and review it -- a shift that changes the nature of the output itself, not just the org chart.

A true 'AI factory' requires multiple layers to function, and one of the core foundational ones is the context layer -- what she calls the organization's 'truth layer.' Without it, AI-native operations don't have a stable foundation to build on.

Groupon's engineering team democratized data access through MCPs, shared frameworks, connectors, and training -- opening the door for people traditionally outside engineering, like salespeople, to build their own dashboards and lead-research tools. The conversations she's having with non-engineers today look completely different than they did just six months ago.

On tooling, the organization started with Cursor but now leans heavily on Claude Co-work, Claude Design, and Claude Code. Training resources for learning these tools are provided free, and she emphasizes that the barrier to entry is effectively zero.

A comprehensive product-capability analysis that used to be an annual exercise now takes just days once data sources are connected, thanks to Claude Co-work -- enabling a research-backed roadmap to come together far faster than before.

Her closing message, repeated for emphasis: become a builder. Don't delegate your understanding of AI to your team -- sit down, roll up your sleeves, and start directing the agents yourself. That path, she argues, advances you further than a team working on the same problem for two years would. Getting started with a coding agent is, in her view, genuinely easy for anyone.