Descriptions:
Jia Wu, deployed engineering lead at Cognition, delivers a conference talk on how the company’s forward deployed engineering motion shapes Devin’s real-world enterprise adoption. Wu provides candid context on Devin’s early reception in 2024 — when it scored 13% on SWE-bench and was dismissed as a curiosity — versus its current trajectory, noting that Cognition’s internal teams shipped nearly an order of magnitude more quality pull requests over six months by deploying their own agent across the organization.
The talk frames forward deployed engineering as a systematic effort to maximize the overlap between Cognition’s product capabilities and the problems enterprises actually face. FDEs are expected to deeply understand a customer’s domain — from feature development and code review to deployment and maintenance — and translate field learnings back into product roadmap decisions, functioning as a high-fidelity feedback loop between customers and the product team.
Wu also outlines the hiring profile for FDEs at Cognition: T-shaped individuals who may come from product management, founding, or software engineering backgrounds, but must carry deep technical spikes. He notes that as the cost of software development falls, FDEs increasingly need product design intuition, not just implementation skill. Wu briefly references his own background from the Windsurf acquisition, contextualizing Cognition’s growing product surface area across CLI, IDE, and cloud agent interfaces.
📺 Source: AI Engineer · Published July 28, 2026
🏷️ Format: Keynote Launch
