An Airy Spirit 0.1.435 is out:
• backups on schedule again on laptops that sleep
• a broken settings file no longer leaks secrets
• cert renewal no longer stops silently
• new: project tag for the wakeup hook
https://t.co/BF3MRUDMgS
#AIAgents#MCP
People, places, organisations - found by GLiNER (Zaratiana et al., 2023). Zero-shot: types are given at run time, not baked in by training.
Under it, a rule layer that needs no model finds dates, emails and links even with the model off.
https://t.co/5Zcoalj9BJ
#MCP#LocalFirst
An Airy Spirit 0.1.433 is out — a security update.
TLS checks no longer trust the compromised eMudhra intermediate certificates (RUSTSEC-2026-0381/0382/0383). Docker images ship the fixed ca-certificates. Your memory stays put.
brew upgrade anairyspirit
https://t.co/KYotObdYMu
#MCP #LocalFirst
Fast search gives a shortlist; a reranker reads each candidate with the question and picks what actually answers it (Nogueira & Cho, 2019).
An Airy Spirit uses bge-reranker-v2-m3, locally. Glance at the pile, then read the few that look right.
https://t.co/5Zcoalj9BJ
#Rust
Two retrievers, two scores on different scales. How to merge?
Reciprocal Rank Fusion (Cormack et al., 2009) ignores the scores and uses only positions. Almost embarrassingly simple - and it beat far fancier methods. We use it.
https://t.co/5ZcoaliBMb
#AI#Memory#AIagents#MCP
Vector search smears rare names, acronyms and exact phrases. Ask for "INV-2291" and you get every invoice that feels similar.
BM25 (Robertson, 1994) doesn't. So An Airy Spirit runs both - each catches what the other misses.
https://t.co/5Zcoalj9BJ
#AI#Memory#AIagents#MCP
"The place with the terrible coffee" should find "café near the station, coffee was awful".
That takes meaning, not word overlap. An Airy Spirit uses BGE-M3 (BAAI, MIT): multilingual vectors of meaning, run locally on your machine.
https://t.co/5Zcoalj9BJ
#AI#Memory#MCP#Rust
"What did I decide last Tuesday?" - when it happened ≠ when it was written down.
An Airy Spirit stores both timestamps for every memory (an idea from temporal databases, Snodgrass 1995), so "last Tuesday" means the world, not the write log.
https://t.co/5Zcoalj9BJ
#AI#Memory
Anairyspirit 0.1.410: local memory for AI agents.
��� Chinese — search, time phrases, tags
🌙 Episodic memory — only what survives sleep stays
📌 Intentions — your agent's plans outlive the session
https://t.co/JUeK7bepBl
#AIAgents #LocalFirst #MCP #AI #中文
ACT-R: memories you use stay reachable, unused ones fade. 40 years of evidence.
We implemented it from the paper, measured it — and on our benchmark it ranked worse. So it ships, switched off.
Not a verdict on ACT-R. A verdict on this workload.
https://t.co/5ZcoaliBMb
#AI
"Cells that fire together wire together" — Hebb, 1949.
In An Airy Spirit, memories recalled together can grow a link even with zero words in common. Similarity can't see that; use can.
An option, off by default.
https://t.co/5ZcoaliBMb
#AI#Memory#AIagents#MCP#LocalFirst
Memories get organised during sleep - the brain replays, structure forms.
An Airy Spirit has a background "sleep" phase for the same job: duplicates merge, links form, topics emerge. You write things as they come; the graph tidies itself later.
https://t.co/5Zcoalj9BJ
Say "fire truck" and "siren" comes to mind on its own. That's spreading activation (Collins & Loftus, 1975).
In An Airy Spirit activation flows through the memory graph, so context surfaces what the query never mentioned. It's what makes memory, not an index.
Most "AI memory" is a vector DB with a nicer name.
An Airy Spirit is modelled on how remembering works, not how search works. Every idea in it has a paper behind it.
Starting tomorrow: one idea a day, who it came from, why we use it.
https://t.co/5Zcoalj9BJ
Not a single call to a large language model — not to store, not to search, not to rank. Public build, default settings.
What these numbers don’t show, we say too: they measure search, not what the graph does beyond it.
https://t.co/5ZcoaliBMb
Memory for agents, built by an agent with memory.
An Airy Spirit is made by two: a person and an AI agent. The person decides what the system is, how it is checked and when it is ready. The agent writes the code — not one line by hand in six months.
LongMemEval, 470 questions — the right past conversation in the top five:
overall 98.5%
facts the user stated 100%
what the assistant said 100%
knowledge that changed 100%
across several sessions 99.2%
reasoning about time 97.6%
preferences 90%