pavelignatov.com

Projects

Personal products and experiments: what I’m building, the decisions behind them, and what remains unfinished.

XRay

From screenshots to a record I can check

XRay brings together mobile-game market signals. In the LiveOps collection experiment, I directed agents to build a workflow that observes screens and stores structured, dated findings with their source captures. I chose the responsibilities and acceptance criteria; the difficult part was catching missing observations, not generating more text.

Scope & limits

The collection article describes a Royal Match experiment, not complete coverage of every game or a measured accuracy rate. The public product separates available market and ad tools from work still in development.

Read the engineering story →

LiveLoop

Make the assumptions visible

I built LiveLoop to explore how resource flows interact before changing a game economy. A node-based model, spreadsheet bindings, and side-by-side variants let me examine a tuning decision without copying an entire model. The linked Last War walkthrough shows the models and the assumptions behind their results.

Scope & limits

A simulation describes the model we give it. It can expose trade-offs, but it does not establish real-player conversion, retention, or revenue. Those need validation outside the model.

Read the engineering story →

My Name Is Pixel

Fewer handoffs, clearer ownership

Pixel is a Unity game in development and my test bed for AI-assisted production. I set the direction and judge the result; agents work on implementation, models, and interface. When a mandatory chain of specialist reviewers slowed progress, I reorganized work around a contributor responsible for a complete result, followed by independent review.

Scope & limits

This is a prototype, not a released game. The development diary distinguishes concept art from game captures and does not claim measured productivity gains.

Read the engineering story →