The Hybrid AI-Horizon Strategy
by Inbal Shani • Chief Product Officer at GitHub
Inbal is the CPO at GitHub, overseeing the product vision for the world's leading developer platform. Previously, she held leadership roles at AWS and Microsoft, and started her career as a senior software developer in Amazon Robotics.
🎙️ Episode Context
Inbal Shani discusses the transformative impact of AI on software development, debunking the myth that AI will replace engineers while highlighting the shift toward system-level thinking. She breaks down GitHub's internal "dogfooding" culture, the metrics that actually matter for AI productivity (beyond just speed), and the future of Copilot as a collaboration layer.
Problem It Solves
Balancing immediate product needs with long-term disruptive innovation without turning research labs into isolated 'ivory towers'.
Framework Overview
Separating product execution from pure research (GitHub Next) while maintaining tight synergy. This allows exploring 'Horizon 3' ideas (3-5 years out) like Copilot Workspace, while product teams focus on 'Horizon 1' execution.
🧠 Framework Structure
Dedicated Research Bandwidth: Create ...
Make it Real: Research cannot just wr...
Fail Forward: Accept that not every e...
Symbiotic Flow: Insights flow from cu...
Niche + General: Prepare for a future...
When to Use
For mature product organizations looking to innovate beyond incremental feature updates.
Common Mistakes
Using research teams for tactical 'Horizon 1' work (urgent fixes), or letting them become too academic with no path to production.
Real World Example
Copilot started as an experiment in the GitHub Next research team in 2020 before becoming a core product. Now they are working on Copilot Workspace based on the same model.
If you try to structure innovation, you're losing that organic spark... GitHub Next's job is to invent the future.
— Inbal Shani