The R&D-to-Production Transfer Protocol
by Ryan J. Salva • VP of Product at GitHub
Ryan is a product leader with a non-traditional background in Philosophy and English, focused on how humans communicate and create. Previously a Principal PM Manager at Microsoft working on Developer Tools and Azure DevOps, he currently leads product at GitHub, overseeing major initiatives like Codespaces, Actions, and the incubation and launch of GitHub Copilot.
🎙️ Episode Context
In this episode, Ryan J. Salva discusses the fascinating origin story of GitHub Copilot, detailing how it evolved from an accidental discovery to a transformative AI product. He shares deep insights on structuring innovation within large organizations, specifically how to transition 'moonshot' projects from R&D labs to scalable production teams, and explores the ethical challenges of building AI pair programmers.
Problem It Solves
How to take high-risk, high-ambiguity 'moonshot' prototypes and successfully scale them into stable, operational products without losing the original vision.
Framework Overview
A structured approach to graduating projects from an innovation lab (GitHub Next) to the core product organization (EPD). It focuses on rigorous personnel management and roadmap ownership transfer to ensure continuity and operational excellence.
🧠 Framework Structure
Ring-fence the R&D team: Protect rese...
Embed Researchers: Move key researche...
Replacement-in-Seat Rule: Researchers...
Transfer Roadmap Ownership: The opera...
When to Use
When incubating Horizon 2 or 3 projects (3-5 years out) that show strong customer signal and need to graduate to general availability.
Common Mistakes
Pulling researchers out too early before knowledge transfer is complete, or letting the R&D team dictate the long-term product roadmap.
Real World Example
GitHub moved researchers from the 'GitHub Next' team into the Copilot EPD squad during the technical preview. They only returned to the lab after new engineers were hired and trained to maintain the AI model and infrastructure.
The criteria for moving researchers back into their R&D team... can't be based on a calendar. It needs to be based on a replacement in seat.
— Ryan J. Salva