Founder and product lead
Gatok AI
An AI assistant over marketplace seller business data.
Constraint
Sellers drown in data and still cannot decide
Key decisions
Keep the assistant close to real seller jobs — sales health, product quality, margin pressure, ad performance, customer questions, claims and returns — rather than building a general analytics chat.
What exists today
—
What exists today
- In active development as a founder-led product.
- Product scope is defined around specific seller jobs rather than general analytics.
- No public results to report yet — this page will be updated when there are.
Context
Mercado Libre sellers operate across commercial, advertising, operational and reputation data. The data exists; turning it into a decision is the expensive part.
This is a founder-led, early-stage product that I am building myself.
The constraint
The value is not in aggregating data — dashboards already do that. It is in answering the question behind the data well enough that a seller will act on it, which makes grounding and trust product requirements rather than technical details.
My role
I cover product strategy, AI interaction design, seller workflows, data-quality questions and monetization direction, and I build the product hands-on.
Key decisions
Keep the assistant close to real seller jobs — sales health, product quality, margin pressure, ad performance, customer questions, claims and returns — rather than building a general analytics chat.
Treat trust as part of the feature: answers have to be grounded in the seller's own business data and clear enough to use in a daily decision.
What I took from it
In seller products, usefulness depends on the quality of the question almost as much as the quality of the answer. The interface has to help people ask better business questions.