Summary
Sunny Rekhi, CTO of Forward Deployed Engineering at Decagon, describes how an AI-native agentic company structures its customer deployment function at the AI Engineer conference. Decagon builds 24/7 AI customer service agents — replacing traditional IVR phone trees and multi-day email queues — and has scaled from 50 to 500 employees in a single year while deploying across verticals from financial institutions to consumer tech brands including Hertz.
Rekhi outlines two distinct FDE tracks at Decagon: agent builders, who configure the AI brain for each enterprise (defining instructions, brand tone, escalation thresholds, and omnichannel behavior largely through the product UI), and agent software engineers, who translate enterprise-specific product requests into generalized platform capabilities. He argues that with AI coding tools now highly capable, the scarce skill is no longer implementation speed — it is exercising restraint, ensuring that a one-off customer request is architected to benefit the next ten customers rather than producing a brittle tangle of prompts and patches.
A key theme is up-front requirements discipline: Rekhi explains that working with Fortune 20 enterprises requires locking down success metrics, channel scope, and expected outcomes in writing before any build begins — a practice that has become more important, not less, as AI coding accelerates implementation. The talk offers a detailed operational blueprint for any agentic AI company navigating the tension between rapid enterprise customization and maintaining a clean, scalable product foundation.
📺 Source: AI Engineer · Published July 28, 2026
🏷️ Format: Keynote Launch
