I built a better, human like memory, for Agents
emson
11 points
1 comment
March 29, 2026
Related Discussions
Found 5 related stories in 46.1ms across 3,471 title embeddings via pgvector HNSW
- Show HN: NERDs – Entity-centered long-term memory for LLM agents tdaltonc · 13 pts · March 06, 2026 · 56% similar
- Custom programming languages make agents good matsur · 17 pts · March 12, 2026 · 50% similar
- Agents that run while I sleep aray07 · 288 pts · March 10, 2026 · 48% similar
- Show HN: I built a sub-500ms latency voice agent from scratch nicktikhonov · 289 pts · March 02, 2026 · 47% similar
- Go hard on agents, not on your filesystem mazieres · 196 pts · March 28, 2026 · 47% similar
Discussion Highlights (1 comments)
slowmovintarget
Love the idea. Please protect your supply-chain. Python projects in the wild, especially as relates to agents, give me the willies these days with the GitHub Actions compromise that's been rolling through the OpenClaw dependencies. (Last one I saw was Telnyx.) How does this blend with something like wedow/ticket, or beads? Those are "short term" in-repo task lists that help shape context. I've been thinking that adding a good long-term memory was the next step. I'll have to kick the tires tomorrow.