Deep Dives
Architecture, engineering, and benchmarks
The Case Against External Vector DBs for Agent Memory
Most 'add memory to your agent' tutorials start with installing a vector database. They shouldn't. Why vector DBs are the wrong default for agent memory.
How Hindsight Scales
A design analysis of how Hindsight's memory operations scale with data volume — what costs grow, what stays bounded, and why.
The Missing Layer in Every Agent Harness
Modern agent harnesses ship with tools, MCP, and IDE integrations — but no memory. Why that's the missing layer, and how harnesses are starting to fix it.
Your Agent Is Not Forgetful. It Was Never Given a Memory.
Why agents seem forgetful, and why memory is different from context windows and retrieval. How Hindsight adds long-term memory to agents.
How We Built a 4-Way Hybrid Search System That Actually Runs in Parallel
Sequential async queries were killing our retrieval latency. Here's how we built a true 4-way parallel hybrid search system with asyncio and RRF fusion — then evolved it further with connection sharing, cross-encoder reranking, and multiplicative boost scoring.
Agent Memory Benchmark: A Manifesto
Agent Memory Benchmark: A Manifesto
How We Built Disposition-Aware Agents That Actually Think Differently
Hindsight — Disposition-Aware Agents
How We Built Time-Aware Spreading Activation for Memory Graphs
Hindsight — Stories, Not Rows
How We Solved Memory Conflicts in Hindsight
Learn how Hindsight handles contradictory information by tracking temporal evolution and preserving history in its memory consolidation system.