Ditching JSON & SQL Friction: Designing an Object-Shaped Memory Engine (Part 2)

<p>Originally published on my blog:<br /> <a href="https://nicolai.hashnode.dev/i-built-a-typescript-object-graph-database-because-i-got-tired-of-flattening-everything"

<p>Most AI agents have amnesia.</p> <p>You close the chat window. They forget everything. Every session starts from zero. Every context window fills up and dumps the earliest

<p><em>Building Ferrous Network exposed the limits of general-purpose databases. Here's why I'm writing GraniteDB from scratch.</em></p> <p>Hey folks, Altug here — founder of

<h2> The Problem We Were Actually Solving </h2> <p>The treasure-hunt server receives 50 MB/s of dynamic map events—player moves, loot spawns, fog-of-war reveals—and must broadcast

<p>Offline reads are relatively straightforward. Put data in SQLite, query it locally, and render the result. Offline writes are where things get uncomfortable.</p> <p>What

<p><strong>TL;DR:</strong> I built the same browser agent twice — once with 500 lines of Python, once with 7 lines of JSON. The second one took 5 minutes. The agent harness layer

<blockquote> <p>The pager went off at 2:14 AM.</p> </blockquote> <p>If you’ve spent any time operating distributed systems, you know that sound. It doesn’t just wake you up; it

<p>Most agent frameworks treat memory as an afterthought. They give your agent tools, prompts, and orchestration patterns — but when you restart the conversation, everything