frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

AIngram – SQLite shared reasoning memory for agent loops, no API keys

https://github.com/bozbuilds/AIngram
2•bozbuilds•1h ago

Comments

bozbuilds•1h ago
Aingram was created to be a local-first memory layer for AI agents/loops. The original intent was for agent memory to use with Karpathy's autoresearch loop, but it's ended up being very capable in general. It stores structured reasoning that will persist across your sessions, which allows separate agents to coordinate tasks, as well as piggy-back off of each other's research loops.

There's just one SQLite file for storage. There's no server, no API, no cloud dependencies. The retrieval mechanism uses a few different systems: FTS5 for keyword matching, sqlite-vec for semantic similarity using nomic-embed-text-v1.5, a QJL two-pass vector compression layer which helps to keep latency manageable for larger databases, and knowledge graph traversal (recursive CTEs) if you want entity-linked results. Those signals are processed via reciprocal ranked fusion, re-ranked, and then give the search results. Everything in the pipeline runs locally.

Benchmarked with LongMemEval, the oracle split gave a recall@3 of 1.000. On LongMemEval S split, recall@10 was 0.955. Median retrieval latency was 22ms/27ms on oracle/s splits respectively, on a laptop with an RTX 4060. The library has an Apache 2.0 license. There's also a 'Pro' tier in development with added optimization features. Let me know what you think!

pip install aingram

GitHub: [https://github.com/bozbuilds/AIngram] | [https://aingram.dev]

bozbuilds•1h ago
A bit of background on why I built this, rather than extending an existing library:

Mem0, Letta, Zep, etc, all target memory as 'personalization' or 'personal context' — they store user preferences, conversation history, individual agent state. None of them use reasoning structure, or are designed for multi-agent memory sharing. The problem I tried to solve was that my agent loops know what changed, but not why it worked or failed.

The retrieval architecture ended up being an interesting engineering problem. Three signals (FTS5, QJL-accelerated sqlite-vec cosine similarity, knowledge graph traversal) are each fast independently, but are for different things. FTS5 is fast and exact. Vector search handles semantic similarity. The knowledge graph surfaces entries connected to entities, even when there's no semantic overlap. RRF fuses the three ranked lists. This hybrid consistently outperforms any single signal on recall quality.

Happy to go deep on any of the architectural decisions.

Top Medical Schools Next Higher Ed Battleground

https://www.bloomberg.com/news/articles/2026-04-08/trump-targets-top-medical-schools-as-next-high...
1•petethomas•2m ago•0 comments

Southwest Airlines imposing new rules on portable chargers on flights

https://www.livenowfox.com/news/southwest-airlines-imposing-new-rules-portable-chargers-flights
1•herecomesthepre•3m ago•0 comments

Show HN: Best of WebRTC Projects

https://github.com/webRTCv1/Best-of-webRTC
1•bestofwebrtc•3m ago•0 comments

Stop using "GitOps" to sell your products

https://lepape.me/stop-using-gitops-to-sell-your-products/
2•FrancoisLp•4m ago•0 comments

Show HN: LLMtary – Local LLM Red-Teaming Tool

https://www.llmtary.com/
1•chetstriker•4m ago•0 comments

How much of success is luck?

https://longform.asmartbear.com/lucky/
2•thm•4m ago•0 comments

"New Sages Unrivalled" – On Mythos

https://www.hyperdimensional.co/p/new-sages-unrivalled
1•speckx•4m ago•0 comments

Router compromise leads to DNS hijacking and adversary-in-the-middle attacks

https://www.microsoft.com/en-us/security/blog/2026/04/07/soho-router-compromise-leads-to-dns-hija...
1•mariuz•4m ago•0 comments

Optimizing our inference back end with custom load balancing

https://www.photoroom.com/inside-photoroom/optimizing-our-inference-backend-with-custom-load-bala...
1•ea016•4m ago•0 comments

Airport Security Wait Times App

https://lufthaven.app/download
1•atharvavaidya•6m ago•0 comments

No big trucks for little roads: American OEMs say EU is blocking imports

https://arstechnica.com/cars/2026/04/no-f-150-in-france-us-automakers-complain-the-eu-blocks-big-...
1•voxadam•6m ago•0 comments

I'm 16 and built a desktop app that controls your PC autonomously

https://www.indiehackers.com/post/im-16-and-built-a-desktop-app-that-controls-your-pc-autonomousl...
1•yousefsaleh-dev•6m ago•0 comments

Muse Spark – Meta Superintelligence Labs

https://meta.ai/
3•snowman647•7m ago•0 comments

Post-Quantum Cryptography for the PHP Community

https://paragonie.com/blog/2026/04/post-quantum-cryptography-for-php-community
1•CiPHPerCoder•7m ago•0 comments

Muse Spark: MSL's First Model, Purpose-Built to Prioritize People

https://about.fb.com/news/2026/04/introducing-muse-spark-meta-superintelligence-labs-first-model-...
3•ilreb•8m ago•1 comments

'Non-survivable': heatwaves are breaching human limits

https://www.theguardian.com/environment/2026/apr/08/extreme-weather-heatwaves-breaching-human-sur...
4•ljf•8m ago•0 comments

How much do Tokens Cost? [video]

https://www.youtube.com/watch?v=u7jO5slUiUw
1•ankit70•11m ago•0 comments

Muse Spark: Scaling Towards Personal Superintelligence

https://ai.meta.com/blog/introducing-muse-spark-msl/?_fb_noscript=1
5•chabons•11m ago•2 comments

Japan relaxes privacy laws to make itself the 'easiest country to develop AI'

https://www.theregister.com/2026/04/08/japan_privacy_law_changes_ai/
2•neitsab•11m ago•0 comments

Show HN: SpoonBaboon – From "I want to make X" to a guided cooking session

https://www.spoonbaboon.com
1•Averkin•11m ago•0 comments

Obsidian will receive the 0.0.3-rc3 brief. You can see and follow clicking here

https://github.com/roading-os/Obsidian-Kernel
1•oretuff•12m ago•0 comments

The twilight of America's sky knights – Fighter jets are an anachronism

https://unherd.com/2026/04/the-twilight-of-americas-sky-knights/
1•voxleone•13m ago•0 comments

Anthropic Just Handed Apache $1.5M to Secure the Open Source Stack AI Depends On

https://itsfoss.com/news/anthropic-apache-software-foundation-donation/
2•speckx•14m ago•2 comments

Omid Teimory

https://omidteimory.com
1•omid2007hope•15m ago•0 comments

Why your new computer is slower than your old computer [video]

https://www.youtube.com/watch?v=t992ul_IKtc
1•ernesto905•17m ago•0 comments

The Privacy Black Hole: Third-Party Doctrine

https://incognitocat.me/the-privacy-black-hole-third-party-doctrine/
1•abnercoimbre•17m ago•0 comments

Ferretlog: Git log for your Claude Code agent runs

https://github.com/eitanlebras/ferretlog
3•eitanlebras•18m ago•2 comments

Show HN: India Trade CLI

https://github.com/hopit-ai/india-trade-cli
7•ArchieIndian•19m ago•0 comments

FFmpeg maintainers thank Anthropic for Mythos patches

https://xcancel.com/FFmpeg/status/2041595801483264002
4•hmokiguess•20m ago•0 comments

Legibility Is Ruining You

https://jimmyhmiller.com/legibility-is-ruining-you
2•jimmyhmiller•20m ago•1 comments