Walking and Talking in the Woods with AI: The Future of Untethered Software Development
Walking and Talking in the Woods with AI: The Future of Untethered Software Development
First, a heartfelt thank you to DevSecCon and Snyk.io for hosting me and giving me the chance to share this vision. The keynote was delivered as a 32‑minute short film that demonstrates everything discussed in this post—from voice-driven development on mountain trails to orchestrating AI agents in production. Watch the full film below to see the complete workflow in action.
Table of contents
The Experiment: Trading the Desk for the Trail
Six months ago, I made a decision that changed everything. I stopped asking how to squeeze more productivity out of new tools while sitting at the same desk, and started asking a different question entirely: what if these tools could help me leave the desk behind? The experiment was deceptively simple to articulate but surprisingly difficult to execute—I wanted to keep shipping secure, tested, production code while spending half my days on mountain trails. What I discovered was that speed requires safety, and that the real power of AI comes from orchestrating both artificial and human intelligence so you can do deep thinking wherever you think best, while background agents safely push well‑scoped changes through a hardened CI/CD pipeline.
I first explored this concept in depth in my article Walking and talking with AI in the woods, which laid the groundwork for this approach.
The unlock came from weaving three essential elements into a tight, continuous loop. First, I needed voice to become my primary interface. With reliable dictation that actually understands technical language, filenames, and identifiers, I found I could speak code at roughly 179 words per minute. That speed matters not because it's impressive on paper, but because it preserves something precious: flow. Ideas travel straight from mind to artifact without the friction of a keyboard acting as an intermediary. I've learned to solve architecture problems around mile three of a trail run, narrating constraints and trade‑offs out loud while the rhythm of movement and fresh air clear my head in ways that no amount of staring at a screen ever could.
I use WisprFlow for high-accuracy voice dictation. Read my full review to learn how it enables 179 WPM development speed.
The second piece of the puzzle was letting background agents pick up those spoken briefs and do real work inside ephemeral, isolated environments. Each task gets its own branch, its own logs, its own previews and diffs. This separation is crucial—it lets machines methodically chew through well‑scoped chores like maintenance, refactors, and visual tweaks, while I stay focused on the things that actually require human judgment: defining goals, weighing constraints, and making strategic decisions. It's orchestration, not micromanagement, and the distinction matters enormously.
But none of this works without the third element: a rigorously protected CI/CD lane. The pipeline must be hardened end‑to‑end, with pre‑commit hooks catching secrets before they ever leave your machine, dependency scans blocking vulnerable states before they merge, and non‑negotiable lint, test, and build gates enforcing quality at every step. Every change is auditable, every decision traceable. That's what actually enables speed—when the path is safe, velocity doesn't erode trust. Instead of slowing down to second‑guess every change, you move confidently because the rails are strong enough to keep you on track.
The Principles That Make It Work
These three technical pieces rest on a foundation of principles that guide how I think about the work itself. The first principle is that speed requires safety. Hardening the lane with pre‑commit secret scanning, CI re‑scans, dependency checks, and gates that simply won't let broken code through creates a path you can trust. When those rails are strong, you can move quickly without the constant fear of silent regressions or accidentally leaking credentials. It's not about moving fast and breaking things—it's about building a system that lets you move fast because it's nearly impossible to break things.
The second principle is to think where you think best. Voice‑driven workflows and mobile check‑ins fundamentally decouple cognition from the keyboard. You can do diffuse‑mode thinking on walks—learning, designing, deciding—then return to your desk to review diffs and integrate the results. The beautiful part is that momentum never stops, because agents keep the code moving forward while you're away. Your mind gets the space it needs to work through problems at a different level, and the implementation keeps pace.
The third principle ties it all together: orchestrate, don't micromanage. Machines excel at well‑scoped, testable work with clear success criteria. Humans excel at setting targets, evaluating trade‑offs, and exercising judgment in ambiguous situations. The key is to treat agents like capable teammates—give them crisp briefs with firm acceptance criteria, then supervise through previews and diffs rather than watching over their shoulder at every keystroke.
Conclusion
The mindset shift is both simple and powerful: orchestrate, don't micromanage. Do your best thinking where you think best, whether that's on a mountain trail or in a quiet room. Let background agents turn spoken intent into production‑grade code, and let a hardened CI/CD path keep you honest and fast. The future of software development isn't about sitting in front of a screen for longer hours—it's about building systems intelligent enough that we can step away from our desks without the work stopping.
Thanks again to DevSecCon and Snyk.io for the platform, and for championing a future where great software isn't tethered to a chair. The tools exist now to make this real. The only question is whether we're willing to reimagine what working looks like.