Blog

Thinking in Public

Architecture decisions, product thinking, and the mechanics of making AI search actually work for e-commerce brands.

Architecture2026-02-175 min

Why Agentic UX Requires Separation Architecture

Everyone is building AI dashboards. Nobody is building AI systems that safely execute. Here's why the future of catalog optimization demands separation between agents that think and engines that act.

Read post
Architecture2026-02-175 min

The Architecture Behind EKOM's AI Search Agent

A structural walkthrough of how EKOM's system actually works — from catalog sync to deployed patch. Four layers, two agents, one rule: agents propose, engines execute.

Read post
Architecture2026-02-164 min

Why Do We Need Our Own Canonical Schema?

Every platform stores product data differently. Without a canonical schema, your AI agent is just guessing. Here's why normalization is the foundation of catalog intelligence.

Read post
Architecture2026-02-153 min

Why Can't We Just Use the LLM Directly?

LLMs are powerful, but pointing GPT-4 at your catalog and saying 'fix it' produces unreliable, unauditable output. Here's why agents need structure.

Read post
Engineering2026-02-143 min

Why Every Write Is a Patch

EKOM never edits production data directly. Every change is a versioned patch with a diff, reason code, risk level, and rollback pointer. Here's why.

Read post
Product2026-02-133 min

There Is No Google Analytics for LLMs

AI platforms don't share analytics. Visibility measurement is synthetic and directional. Here's what that means and why it's still useful.

Read post
Product2026-02-123 min

Why Agents, Not Dashboards?

Dashboards show you what's broken. Agents fix it. Here's why EKOM shifted from monitoring to action.

Read post