Blog
Architecture deep-dives, benchmark results, and thoughts on the future of agent memory.
Why Graph Memory for AI Agents Shouldn't Cost $249/Month
Most memory systems charge $249/month to unlock graph features. We're changing that. Here's why the LLM extraction tax makes graph memory expensive — and how GraphPalace eliminates it.
Pheromone-Guided Retrieval: How Ants Inspired Our Memory Architecture
Stigmergic systems — like ant colonies finding food — use environmental signals to coordinate without central control. We applied the same principle to agent memory retrieval. Coming soon.
Benchmarking Memory: $0 Ingest vs LLM-Extracted Graphs
A reproducible head-to-head on LOCOMO and LongMemEval. How a zero-LLM Rust/LadybugDB engine delivers competitive recall at $0 ingest cost — versus LLM-extracted graphs that bill tokens on every write. Coming Month 2.
Stay updated
New posts on graph memory architecture, benchmark results, and product updates. Low volume, high signal.
No spam. Unsubscribe anytime.