Natalie Meurer leads agent engineering at Sierra, building teams that turn AI agents into reliable customer-facing software. She has helped define agent engineering as a discipline accountable for enterprise deployments and measurable customer outcomes, drawing on earlier work in privacy engineering, government software, and startup leadership.
Meurer studied Science, Technology, and International Affairs at Georgetown University, where her honors thesis explored the importance of deleting personal information. She subsequently learned to code and worked at Palantir from 2016 to 2021, spanning civil-liberties software, infrastructure, forward-deployed engineering, and privacy engineering for its government platform. Her customers included law-enforcement and defense agencies, and she co-invented a patented audit-logging system for reviewing activity involving sensitive data.
Her software projects include a naive Bayes classifier with entropy-based preprocessing and cross-validation, and Schedulize, an application that uses a maximum-flow algorithm to assign employees shifts.
After Palantir, Meurer earned an MBA from Stanford Graduate School of Business in 2023. She co-founded Probably Something with Beth Nakamura and served as chief executive, developing Lifeboat to help people navigate cryptocurrency communities and digital ownership. After participating in Variant’s 2023 founder fellowship, the founders closed the company and both joined Sierra.
Meurer joined Sierra’s deployment team and, in 2024, defined agent engineering as customer-accountable applied AI: building agents that reflect company policies, integrate with operational systems, and perform reliably outside demonstrations. She also co-authored an account of Sierra University, which teaches customer teams to design and deploy production-ready agents.
- FDE vintages: Forward-deployed engineering accumulated distinct responsibilities over time: maintaining on-premises deployments, integrating customer data, modeling information through Palantir’s Ontology, building applications, and enabling customers to operate the platform themselves. Meurer’s history of the discipline explains why identical job titles can conceal radically different expertise.
- Customer accountability when code becomes cheap: Coding agents make software easier to produce, increasing the importance of understanding customer problems, translating field experience into product decisions, and owning whether deployments succeed. Meurer sees product engineering and customer-facing engineering converging around those responsibilities.
- Production agent supervision: Dependable agents require supervisory models, enterprise integrations, application-specific constraints, and deterministic responses when regulated interactions demand exact wording. Reliability comes from the surrounding system, not confidence in a single model.
- Outcome-based pricing: Charging for resolved inquiries or completed transactions ties commercial success directly to deployment quality. For Meurer, engineers must remain close enough to customers to understand, measure, and improve the results their agents deliver.
