One Team’s Journey to Using AI Without the Hype
summary
Bottom-up adoption at Octopus lets developers experiment with Copilot, Claude, Cursor, Gemini and others rather than forcing a single tool. Teams generate large portions of code (one repo is 95% generated), see measurable productivity (surveyed users report 3.9 hours/week saved and ~27% merge code from AI), and build product features: a Chrome extension AI assistant, a recovery agent that diagnoses failures via GPT on Azure functions, Rule sync and skills-based guardrails. Hallucinations, bogus APIs and oversized PRs force strict tests, small iterative prompts and instruction files.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 6:03Octopus Deploy tech stack
- 8:29AI experimentation start
- 15:05AI tools and models
- 21:04AI adoption failures
- 26:22Trust in AI code
- 28:07AI workflows and skills
- 37:30AI in product features
- 43:51Recovery agent demos
- 48:07AI Pulse Report findings
- 50:11Future of AI deployments
- 53:20Key takeaways
- 56:04Q&A
related talks
32mBeyond the Hype: A Pragmatic Look at How Developers Actually Use AI Tools
35mBeyond the Hype and Hyperbole
57mAccelerate Your Developer Productivity with AI Embrace the Future Now!
14mDeveloper productivity in a post-AI world
01h 16mAI in Software Development: Boosting Efficiency
41mAI or Not AI? We Don't (Yet) Believe the AI That Replaces Devs!
26mWhen Agile and AI Collide
22mUnlocking AI Powered DevOps Within Your Organization