From Agency to AI: A Non-Linear Path

My path into AI was not a straight line. It started in agency work, moved through web3, detoured into experiments that failed, and eventually landed in the practical, operational reality of shipping AI products for real businesses.

I started Lavon as a digital agency. We built sites, funnels, and brand systems for clients who needed outcomes, not hype. In that world you learn quickly: strategy is only useful if it ships. The agency years taught me how to talk to customers, how to interpret vague goals into clear deliverables, and how to build repeatable systems under pressure. There is no romanticism in agency work, and that is exactly why it was such good training.

From there I moved into web3. The experience was intense and volatile. I worked on Realm of Karitha and helped manage the Tipsycoin crisis. In that cycle I learned how fast momentum can evaporate when trust collapses. I learned to think about incentives, community psychology, and long-tail operational risk. Most importantly, I learned that shiny narratives are fragile. What lasts is execution and accountability.

The AI pivot came from fatigue with fragile narratives. In 2023 and 2024 I watched many “AI startups” ship demos instead of systems. I wanted to build the unglamorous layer: the workflows that actually help a business make money, save time, or remove friction. That was the real gap. Not capability. Reliability.

That shift pushed me toward building practical AI tools for small businesses. A good example is NextGame.dev. It looks simple on the surface: AI-powered game recommendations. Underneath, it was a sprint in product discipline. Building a product people actually use taught me to focus on outcomes, not features. The wins came from attention to detail and tight feedback loops, not from grand strategy.

Today I serve as Chief Digital Officer at Tile Mart, leading digital transformation across their operations. That is where the AI work became concrete. This is not theory. It is about building systems that help teams move faster, reduce errors, and make decisions with clarity. Sometimes that means building internal tools. Sometimes it means automating painful workflows. Sometimes it means refusing to ship a flashy demo until the pipes are solid.

If you zoom out, the pattern is consistent. Every phase forced a different kind of discipline:

• Agency work taught me to ship under constraints and make outcomes measurable.
• Web3 taught me how fast confidence breaks when execution slips.
• The AI pivot taught me that leverage only matters if it is operationally reliable.

The result is a mindset that favors practical systems over promising narratives. When I build today, I think about three questions:

1. Does this make execution faster or clearer?
2. Can a non-technical team operate it without constant hand-holding?
3. Will this still work when the novelty wears off?

That is the lens I apply to every product I ship. It is also why I am obsessive about workflows, automation, and documentation. These are not cosmetic details. They are the difference between a demo and a durable system.

The “non-linear” path is not an apology. It is an advantage. The agency years taught me empathy. The web3 years taught me resilience. The AI years taught me leverage. The Tile Mart role taught me how to embed all of it into a real business with real constraints.

If you are early in your own path, the lesson is simple: your “detours” are not wasted. They are the parts of the story that will make your work distinct. The industries you think are unrelated are the ones that give you uncommon instincts. The pattern I see now is clear: the fastest builders are not always the best. The builders who understand operations, trust, and accountability are the ones who last.

That is what I am building toward. Not just AI products that ship, but systems that stay useful long after the hype wave moves on.

What I build now

Today my work sits at the intersection of product, automation, and operations. The projects vary, but the pattern is consistent:

• simplify workflows so teams move faster
• reduce manual steps without breaking visibility
• build tooling that survives staff changes and context loss

Some of that is software. Some of it is documentation and process. The key is the same: make the system reliable enough that it does not depend on me being awake at 2am.

The principles that guide me

Over time I have built a small set of rules. They are boring on purpose:

1. If it cannot be operated by someone else, it is not finished.
2. If it cannot be explained in one page, it is too complex.
3. If the “happy path” relies on perfect input, it will fail in the real world.

These rules came from hard lessons, not theory. Agency work taught me that stakeholders change every week. Web3 taught me that trust evaporates fast. AI taught me that you can ship fast, but you still need a reliable loop to keep it alive.

The products that shaped this lens

The pattern showed up across every project I shipped:

• NextGame.dev taught me the value of a tight recommendation loop. Users do not want a catalogue. They want a decision.
• Resolution Tracker taught me that a habit loop must be short and obvious. If the value is not felt in minutes, the app dies.
• Internal tooling taught me that the best UX is often removing steps entirely.

These are small products, but the lessons scale. The system matters more than the interface. The loop matters more than the features.

Why the non‑linear path matters

The non‑linear path gave me range. It exposed me to different incentives, different failure modes, and different types of accountability. That range now shows up as judgment. It lets me see the hidden costs in a plan and the fragile assumptions in a roadmap.

If I had taken a straight path, I might be a better specialist. The non‑linear path made me a better operator. And in product, operations are the difference between a launch and a business.

The Tile Mart lens

Leading digital transformation at Tile Mart made everything concrete. This is not a startup sandbox. It is a real business with real constraints. If a system slows sales, it gets cut. If it saves time, it survives. That pressure keeps the work honest.

The biggest shift was moving from “shipping features” to “shipping outcomes.” That is the lens I apply now.

The story I want to keep telling

I am not trying to be the loudest voice in AI. I am trying to be the one who ships reliable systems. That means fewer demos, more boring checklists, and more time spent tightening the loops that make a business predictable.

That is the real throughline. Every phase of the journey pushed me toward that. I intend to keep walking it.

---

Related Guides

• Melbourne AI Startup Guide — Build, validate, and launch locally
• AI Product Validation Guide — Prove demand first
• Solo Founder AI Stack — Compete with agencies

Related Stories

• From Agency to AI Products — The shorter version of the pivot
• 2025 Wrapped — A year of building in public

Learn More

For the full product path, join the AI Product Building Course.