Neil Stryjski — Product x AI

01 — The Epiphany

A few months ago my gym owner told me he'd tried and failed to get custom software built for his business. My instinct was immediate: help him define the product and make sure a developer builds it right.

Six weeks later, I realized I could have built it myself. Not handed it off. Not found a developer. Built it — concept to active users — myself.

The developer role didn't disappear. It moved from entry point to scale-up partner for a product that's already in the hands of real users. That's a fundamentally different build cycle, and it only works if the Product thinking is rigorous enough to drive the build.

02 — The Argument

AI doesn't have a speed problem. It has a direction problem.

Read the full argument

The tools are fast. The generation is instant. The gap is knowing what to build, why it matters, and whether the output actually solves the right problem. That's not an AI capability — that's Product discipline. And without it, AI velocity is just the fastest way to build the wrong thing.

The Product function has always been the layer that turns ambition into something shippable. Discovery before commitment. Scoping before build. Governance that holds without killing momentum. Those practices don't become less important when AI enters the build cycle — they become the only thing that makes AI's speed count.

The new build cycle demands a Product practitioner who can operate across the full stack — pushing left into technical architecture before a line of code is written, and pushing right into working software that proves the intent rather than just describing it. Not handing off to engineering. Not waiting for a prototype to validate the thinking. Building.

What this demands — counter-intuitively — is a new consensus with engineering. Not a power struggle. Both functions are pushing their boundaries outward simultaneously. Product deeper into the technical. Engineering deeper into the strategic. The new ways of working only hold if both sides author them together.

None of this is context-free.

The LinkedIn feed is full of founders who shipped a product in a weekend, solo builders who've never written a line of code deploying MVPs to live users. That's real — and it's genuinely remarkable. But it describes one specific operating condition: a single person, solving their own problem, carrying the full context, with no alignment cost and no risk surface they can't see.

Scale that to an enterprise and every variable changes. Organizations run by humans have always struggled to absorb digital disruption — not because of technical limitations, but because adoption is a human problem. The new ways of working that AI demands don't arrive fully formed. They have to be negotiated, modelled, and earned across functions that are simultaneously being asked to redefine their own boundaries.

The CEO who just discovered Claude Code isn't wrong to feel the possibility. The job is to channel that energy into the right decisions — not faster outputs, but better bets. That's always been Product's seat. The new build cycle just made the stakes higher and the visibility more uncomfortable.

That's what this practice is.

03 — The Framework

The Product Trio Agentic Process. A structured framework governing how Product, Engineering, and Design collaborate with AI agents across the full development lifecycle.

Brain → Bedrock → Hands. Three phases. Two environments. One source of truth.

Go deeper — The Product Trio Agentic Process

04 — The Proof

The thesis needed a test. Vechelon is it.

Now in live trial with Racer Sportif

Read the build arc

Vechelon is a cycling club management platform — real-time fleet visibility for ride leaders, zero-friction guest entry, and ephemeral data privacy built in from the start. It is in live production trial with an active cycling club. It was built by a product leader who has never written a line of code.

That last sentence is the point.

The Bedrock evolved through three phases in roughly two months. Each Pillar version below is the formally-committed document that drove the next stretch of build.

Foundation and Vision

Feb 2026

Locked the north star before a line of code was written. Original hard constraints — $0 operating cost, zero-friction guest entry, 4-hour GPS purge.

+ ~4 weeks

+ ~4 weeks elapsed

Pivot

Mar 2026

Concentrate on real depth in the member experience first. Mobile branch deferred, desktop-first assumed primacy.

+ ~4 weeks

+ ~4 weeks elapsed

Launch Readiness

Apr 2026

Whether the architecture was actually right. All three surfaces — Admin, Rider Desktop, and the VoC / MT / IA layer — folded into one Bedrock ready for live production use.

Three phases. Two months. One sole practitioner. The velocity story is the rail.

05 — The Practitioner

Fifteen-plus years building product in environments where failure has real consequences — $40M ARR platforms, enterprise telco, nationwide network deployments — builds a specific kind of discipline. That discipline is what makes AI fast in the right direction rather than just fast.

The full story

06 — The Thinking

The argument keeps evolving. Here's where the thinking is documented in the open.