How Intercom Doubles Down on AI
summary
Intercom rebuilds its product and engineering model around Finn, an AI agent launched alongside GPT‑4 that powers a million+ resolutions per week, proprietary models that beat general LLMs, and agent plugins auto‑deployed to every developer. Goals include doubling developer productivity in 12 months, moving 80–90% of PRs to cloud‑code driven workflows, centralizing telemetry and LLM judges, pulling top engineers onto a focused team, and replacing manual QA with recorded sessions that agents parse — results so far are rapid throughput gains and accelerating improvement.
AI-written summary · about
chapters (8)
- 0:00Introduction and AI urgency
- 3:57Intercom’s AI pivot and Finn launch
- 9:16From spades to diggers in software
- 12:49Phase one of AI tool adoption
- 16:43Phase two: Opinionated AI systems
- 26:24Phase three: Machines building machines
- 30:58Leadership lessons in AI adaptation
- 37:02Call to action and future challenges
related talks
49mHow to be an AI First company
18mDispatch from the Future: building an AI-native Company
33mBuilding the Next Generation of Software
56mEnabling teams with AI
11mRuby startup demo – Ryan Sherlock
48mMastering the AI Shift: Adaptive Strategies for Software Evolution
46mRecalibrate - How AI shakes up software team
26mA reality check on AI’s impact on business transformation