How Fraktal came to be?

Fraktal started as something smaller: an AI agent for SEO. Twenty days of building and talking to founders turned that agent into Fraktal.

Chapter 1. Founders struggled to explain their own products.
7 October.
The three of us, Luka, Bojan and Jovan, had talked for months about one problem. Founders build good products, then struggle to say what the product does and who it is for. We decided to start with the part we knew best: getting found on Google and in AI answers. We built an AI agent for it and called it Timothy.
8 to 10 October.
We tested the tools that track what AI engines say about a company. The data was good. Every tool stopped at the numbers, though, and none of them said what to do next.
11 October.
We ran our first free audit for a startup. It showed which buyer questions the startup was losing to its rivals, and what to fix first.
14 to 21 October.
Timothy got a face: a brand, a 3D character and a short launch film.
Chapter 2. We tested it on real startups.
16 and 17 October.
We took ten real B2B startups. We asked six AI engines and Google the buyer questions of each startup, and we stored every answer.
19 to 21 October.
Each founder got a report on how AI engines see their company.
Chapter 3. Every tool we tested gave stale advice.
26 October.
The same flaw showed up in every GTM tool we had tested. Each one gives advice from a snapshot, and nobody checks if that advice is still true next month. We named the company Fraktal and set out to build one brain that knows what works now on every channel.
27 October.
We asked the ten startups' questions again. In ten days, the share of ChatGPT answers that cite big news sites fell from 25% to 2.5%. The first reading was already out of date.
Chapter 4. Founders showed us what they need.
28 October.
Luka needed an interactive demo for a company he runs GTM for. The leading tool, with 3,000 customers, felt clunky, so he built his own with an AI agent in about an hour. It fit the company better. Jovan put it this way: two companies can look alike, but once their own data goes in, their needs are different. So every founder should build their own tool, on current data.
28 October, afternoon.
We spoke with the growth team of a startup that runs dozens of AI agents. They had tried three AI search tools. In their words, insight means nothing until you act on it. They asked for current data their own agents can use, through an API and an MCP.
Today.
Fraktal is a data layer for the next generation of GTM tools. The brain tracks what works on every channel, and the toolkit helps founders build their own GTM tools on top of it.

Team

Who builds Fraktal.

Growth, data systems and buyer psychology.

Luka Čereković
CEO
Has run go-to-market and growth for startups in several countries for eight years. He raised millions in venture capital for the startups he worked with, and generated millions in revenue.
Bojan Stefanović
CTO
Former engineer at Reddit and Google. Labeled AI training data for Mercor and Anthropic. He builds the infrastructure.
Jovan Jovanović
GCD
Has studied how people communicate, and why, for ten years: psychology, neuroscience and what makes a marketing campaign work.