Growth is a research loop.
I spent years inside fast-moving AI products, then years advising founders who couldn't get found — now I write the loops down, ship my own, and open-source what compounds.
I didn't walk in through a consulting deck. I walked in through languages — five years in Russia, five more in Germany — sitting inside other people's words until I understood something most product teams miss: the distance between what you built and what a stranger will type is not a copywriting problem. It's the whole game.
August 2022. I took the operator seat at an AI imaging company the month the category caught fire. Midjourney landed; ChatGPT followed; every roadmap doubled overnight. While engineers raced on models, my race was elsewhere — making sure launch day wasn't the peak, that taxonomy pages kept earning while the team slept, that a shared link didn't die in Slack because the preview looked like a placeholder. I wasn't choosing winners. I was building the rails.
Late 2024. I went independent — same craft, wider lens. Client work taught me how voice, video, agent, and vertical AI teams break differently. Founder calls taught me how the stack actually connects — labs, infra, dev tools, distribution layers — when nobody's performing for a pitch deck. Somewhere in that volume I stopped treating each engagement as a one-off.
Find the signal. Write it down. Ship it as a system.
That's the thread: operator first, advisor second, builder third — packaging what survived contact with reality so the next team doesn't pay the same tuition.
| 100+ AI/SaaS clients since going independent |
200+ founder conversations across the AI/Agent stack |
3 products I ship and run myself |
170+ open-source agent skills in git |
Growth lead · 2022–2024 · in-house products, not client work.
Before I had a studio name, I had a seat inside two global GenAI products — one holding a16z's top-50 list while shipping APIs and video tools faster than most marketing orgs could brief an agency; the other pushing sketch rendering and design workflows into markets that didn't know they wanted AI yet.
That velocity is where I learned the unglamorous truth: most AI products lose on silence, not on model quality. Not client logos below — the training ground.
Three outlets. One habit — turn field notes into something the next person can run.
|
Alignify · where the research lives Every retainer leaves behind a constraint worth remembering — a head term too broad to rank, a share surface that embarrassed the founder, a taxonomy that should have compounded and didn't. Alignify is where those field notes become bilingual playbooks: SEO architecture, tool evaluation, distribution loops written so the next team can execute without me in the room. |
Oginify · where I fixed sharing There is a specific humiliation in watching a beautiful launch die in a group chat because the link preview looks broken. I saw it on repeat across client work — so in June 2026 I shipped my own answer in forty-eight hours: paste a URL, get four on-brand OG cards, no signup theater. The first product that's entirely mine, built with the same vibe-coding stack I now document in public. |
Nova Scientia · where I test going local Brazil doesn't need another translated English blog — it needs a map in Portuguese, with honest comparisons and search intent that actually matches how people look. Nova Scientia is me running the same research discipline in another language: 435+ tools reviewed, localization as strategy shipped in the open, not as a side project slide. |
The loop doesn't stop at publish — it has to survive inside Cursor, Claude Code, and whatever agent stack teams wire up next. These repos are the execution layer: strategy in the sites above, mechanics in git.
|
When agents started reading Markdown for a living, I packaged a decade of growth craft into skills — page types, GEO, ads, analytics — so the playbook travels with the codebase, not in a Notion graveyard. |
The spec I wished existed before Oginify — programmatic OG generation with Satori and resvg, built for agents that need share surfaces without opening Figma. |
The stack notes from shipping Oginify in two days — default configs for Lovable-class builders, plus the moment you outgrow the prototype and need Next.js with real SEO hygiene. |
Client work since going independent — logos for illustration, not endorsements.
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
||
 |
 |
 |
 |
 |
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |