Kyle Daigle, the CEO of GitHub and CMO of Developer for Microsoft, says AI has made him 14 times more productive in his personal coding output. Yes, you read that right: 14x. While many leaders preach the gospel of AI, Daigle actually lives it, transforming his own workflow to automate complex leadership tasks and stay close to the code.

He revealed how AI agents now handle non-technical work, like crafting weekly messaging plans. Instead of starting from scratch, Daigle feeds the AI vast amounts of past data, asking it to analyze what worked, what didn't, and then suggest future actions. This isn't about AI building the next big thing from thin air; it's about its ability to perfectly analyze what came before. Daigle even built an “entire presentation without touching any of it” using this method, a task that would normally eat hours of a leader's week.

“I think what you see there is me like really getting back to coding thanks to AI,” Daigle said, highlighting how this shift lets him remain hands-on. His approach flips the script: instead of focusing on AI’s forward-building capabilities, he emphasizes its strength in retrospective analysis to inform the future.

Key Takeaways

  • Kyle Daigle, CEO of GitHub, increased his personal code commits by an unheard-of 14x by integrating AI into his daily workflow. This lets a high-level leader stay deeply connected to technical work.
  • AI's true power, Daigle argues, lies not just in creating new things but in its near-perfect ability to analyze historical data and find patterns for "retrospection."
  • He uses AI agents to automate complex, non-technical leadership tasks, such as generating weekly messaging plans based on past performance or building full presentations from data.
  • This enables leaders to quickly create functional, data-manipulating apps or systems, helping them regain a technical edge even in management roles. Daigle notes the ability to “make it look humanly bad and you know like and build a little app to like manipulate the data I think is part of like that upside for devs that are now in leadership roles.”
  • The core method behind Daigle's dramatic productivity boost is his "Recursive Loop Backwards" Planning Method.

The Kyle Daigle's "Recursive Loop Backwards" Planning Method

This method emphasizes using AI for retrospective analysis to inform immediate future actions, particularly for non-technical or communication-heavy tasks.

  • Step 1: Retrospection: Go back through the week and tell me what we did, what worked, what didn't work...
  • Step 2: Forward Planning: ...and then tell me in the next, you know, three or four days, what would you tweak based on, you know, this sort of like looking backwards and then looking ahead a little bit.

When This Works (and When It Doesn't)

Daigle finds this method especially valuable for non-technical tasks because large language models excel at finding patterns and pulling insights from past data. They can then apply that retrospection to tweak actions over a short, defined period. This framework shines when you have a history of data or actions to analyze, and your goal is to optimize or iterate on existing processes rather than invent something entirely new.

However, it's less effective for truly greenfield problems where there's no past data to draw from. It also struggles with tasks requiring deep human empathy, nuanced negotiation, or truly creative leaps that AI models can't yet synthesize. Use it to refine and improve, not to conjure breakthrough ideas out of thin air.

What to Do With This

Apply Daigle's "Recursive Loop Backwards" method to your weekly investor updates. Gather the last two months of your product updates, team feedback, and any investor responses. Feed this data to an AI agent, then walk through the framework's steps.

First, for Step 1: Retrospection, ask the AI: “Go back through the last 8 weeks of our product updates, team feedback, and investor responses. Tell me what we announced, what worked well in terms of engagement or feedback, and what didn't land as expected.” Next, for Step 2: Forward Planning, instruct it: "Based on this analysis, tell me what you would tweak in our next product update, scheduled for next Tuesday. Specifically, suggest three changes to make it more impactful or address past issues, focusing on the next three to four days." This process delivers a data-backed, actionable plan for your next communication, letting you iterate faster and with more precision.