VULKN operated from March to July 2026. We wound it down deliberately — what we learned now runs at Sigea →
What we built.
We built and orchestrated production AI agents — operations, distribution, and communications layers — deployed with real enterprise clients.
Agents that ran the day-to-day work inside our clients — end to end, on their own.
What kept long-running agents on task instead of losing the thread halfway through.
The whole paid-growth machine, run by agents that sharpened the loop every day.
The real-time channel our agents spoke through — whatever the client already used.
Agents that studied the market and pinned down the exact customer worth chasing.
Ad creative and UGC-style video, produced at volume — no agency, no bottleneck.
Agents read the numbers and improved the next round, every single day.
Getting all of it live inside real enterprises — and keeping it running.
How it actually worked.
We didn't rent the intelligence. We built the agents and the orchestration ourselves, then ran them inside real businesses — and every deployment taught the next.
Built the core ourselves
We engineered the agents and the orchestration in-house — the operations layer most companies could only rent.
Deployed in production
Not demos. Agents that ran operations, distribution and communications inside real enterprises, end to end.
Learned, then let go
Every deployment taught the next one. It also taught us the limit — and the limit is why this page is an archive.
We ran the distribution.
Building the agents was half of it. On top, we ran the paid-growth machine — research, creative and every ad channel — driven by agents that sharpened the loop every single day.
Meta Ads
TikTok Ads
Google Ads
Instagram
WhatsAppLocked the ICP
Agents studied the market and pinned down the exact ideal customer — who to reach and what made them buy.
Generated the creative
AI-made ad creative and UGC-style video, tuned to that ICP and produced at volume — no agency, no bottleneck.
Launched everywhere
Pushed live across every paid channel we could reach, inside one AI-run sales funnel.
Sharpened the loop
Agents read the numbers and improved the next round — the cycle compounded internally, every single day.
What happened.
Between March and July 2026, Vulkn built and operated production AI agents for enterprises. Our first client was an insurer with MXN 1.4B in annual revenue; clients paid between US$1K and US$13K a month — every deal closed with zero network.
We learned the hard way that optimizing agents for every vertical at once doesn't scale. So we made the call most teams make too late: we wound Vulkn down deliberately, handed our flagship client to a team better built to serve them, and took the lessons with us.
What we learned here runs somewhere else now: Sigea →
— Johan Martínez Ríos, co-founder