News for hackers. Stuff that matters.

kendallgclark

Re:RIPL: Real-time Intelligence When Every Millisecond Counts on Thursday October 22, 2026 @06:17PM (#50201851)
Jev models make agents faster. Hooked control planes steal the difference.

https://pentad.ai/PLRN/031/

Re:The OpenAI proofs are us on Wednesday October 07, 2026 @10:30PM (#50001296)
The part where we don’t understand the proofs!
Re:An agent fleet needs a new kind of OS, not a bigger harness on Monday September 21, 2026 @03:52PM (#49792455)
https://pentad.ai/PLRN/027/ — A harness can’t keep an agent’s future-dated promises. An agent OS can.

Discusses some issues raised in this thread.

PlatypusDB is an agent-native, bitemporal event database: Gleam (OTP) control plane over a Zig data plane, in-process with WunderOS. Custom-built, not assembled, for agentic workloads inside WunderOS. The Merkle WAL is the database. Graph, vector, versioned-tree, image, and analogy views derive from it. Datalog queries and rules read it crisply. VSA resonance reads it fuzzily. The whole database replays deterministically and is fuzzed by a TigerBeetle-style VOPR. PlatypusDB is one write path, two clocks, zero glue code.
Re:The Social Tier: Remembering Who Said What on Wednesday July 08, 2026 @08:44AM (#48831211)
Thanks. Gotta be the most often ignored rule on this site. But still.
Re:Looking for Partner to Build Agent Memory (Zig/Erlang) on Monday March 16, 2026 @08:00AM (#47397823)
Thinking about data in terms of energy basins, frequencies, and SNR is different than what I’ve done before.

But pushing a signal below the noise floor is analogous to tombstoning a tuple in a database.

Re:Looking for Partner to Build Agent Memory (Zig/Erlang) on Friday March 13, 2026 @09:41PM (#47372441)
Thanks for the question.

Inlike vector databases that append embeddings to an HNSW graph, my working memory substrate natively supports mathematical forgetting.

I use a Squelch primitive—a SIMD-parallelized saturating subtraction over an 8-bit probabilistic accumulator.

When an agent finishes a chain-of-thought, we literally subtract the statistical mass of that specific reasoning path out of the 16,384-dimensional superposition.

It intentionally drops the signal back below the Kanerva noise floor, freeing up capacity in the L3 cache without destroying the other superposed facts. It works similarly for episodic memory and procedural memory.

Semantic memory, currently, is for invariants and “the schema” so no “deletions” but this will likely be reworked.

As for retrieval, yeah similarity via Hamming is table stakes. But there’s other stuff too including resonator network factorization and a Datalog variant, too.

We map Datalog semantics directly to VSA using 16,384-dimensional hypervectors.

Instead of relational tables, our EDB consists of 'Pentads' (Subj, Pred, Obj, Context, Lineage) bound together using prime-number circular bit-shifts to encode grammatical roles.

These facts are superposed into an 8-bit probabilistic accumulator.

For the IDB and schema enforcement, we use a 'Warden' actor in Gleam that intercepts state changes in the Tuplespace, validating them against constraints before they ever cross the C-ABI boundary.

When we query the Datalog, there are no B-trees or graph traversals.

We construct a 16k-bit probe and use AVX-512 SIMD to perform Maximum Likelihood Decoding directly against the superposed noise in O(1).

Because standard Datalog struggles with time, we extended the semantics to support LTL+ (Next, Eventually, Always, Until) natively over the vector space.

Our Episodic memory isn't a flat table; it's a strict chronological linked-list of physical accumulators stitched together with Holographic Pointers.

To evaluate temporal modalities like Eventually (◊) or Until (U), we don't use expensive SQL window functions or graph traversals.

The Zig sidecar just follows the Holographic Pointers, performing a O(1) SIMD resonance check at each temporal node.

If the target state resonates out of the noise at, say, Node 5, the LTL query resolves to true.

We execute temporal logic as a recursive physical jump through a non-Euclidean probability space.

Working next week on encoding Agent text into vectors directly without an LLM or SLM to assist.

I hope that helps!

Re:Looking for Partner to Build Agent Memory (Zig/Erlang) on Friday March 13, 2026 @09:21PM (#47372300)
Well probably this recent piece by Kanerva. https://arxiv.org/abs/2503.23608
Re:Looking for Partner to Build Agent Memory (Zig/Erlang) on Friday March 13, 2026 @09:19PM (#47372291)
Thanks. Yes all spaces are crowded.

IMO the value here will be quasi brain-like operations on data that are fast and efficient.

We overuse LLMs which aren’t too fast and very inefficient.

So the value here is being able to support a shift of some workloads from LLM to smart agentic memory.

Re:Looking for Partner to Build Agent Memory (Zig/Erlang) on Thursday March 12, 2026 @04:29PM (#47356610)
State of the art for HDC/VSA? Or for agentic memory?

Looking for Partner to Build Agent Memory (Zig/Erlang)

I’m working on a purpose-built memory platform for autonomous AI agents.Right now, agent memory is stuck between two hohum options: RAG (which loses relational topology) and Graph Databases (which require massive pointer chasing and degrade under heavy recursive reasoning).I'm building an alternative using Vector Symbolic Architecture (Hyperdimensional Computing). By mathematically binding...

( Read More... | 10 comments | 2 points | on HN )
Re:RAG Is a Fancy, Lying Search Engine on Monday June 16, 2025 @09:33PM (#44294945)
Not at all. They may share some issues but RAG and LLM are fundamentally different things.
Re:RAG Is a Fancy, Lying Search Engine on Sunday June 15, 2025 @02:07PM (#44283912)
LOL. No. All me, hater.
Re:RAG Is a Fancy, Lying Search Engine on Sunday June 15, 2025 @02:06PM (#44283907)
Fixed. Thanks.
Re:RAG Is a Fancy, Lying Search Engine on Saturday June 14, 2025 @10:52PM (#44280206)
That might be RAG’s benefit if LLMs were more steerable but they can be stubborn.
Re:Alasdair MacIntyre Had Died on Friday May 23, 2025 @10:31AM (#44073265)
He didn’t write as if he hated liberalism. Maybe he did. But in his work you get deep, principled critique from the basis of epistemology and selfhood.

Lenin wrote like someone who hates liberalism. Stephen Miller gives that vibe from the right, though I doubt he can write anything coherent at all.

Not really competitive with Stardog given our leading LLM integration with Voicebox. 85% pass@1 to exit POV with new customer.
The use case at NASA isn’t even new. We built this precise thing in 2008. All standards-based.

See https://www.w3.org/2001/sw/sweo/public/UseCases/Nasa/ for a public case study.

This work led to Stardog.

Which is used extensively in NASA today—

https://gpdisonline.com/wp-content/uploads/2019/09/StardogNA...

https://www.informationweek.com/machine-learning-ai/stardog-...

Re:MillenniumDB: A graph database engine using domain graphs on Saturday February 01, 2025 @10:48PM (#42905632)
That’s tough. Not sure what that has to do with Stardog. Biggest companies in the world rely on it daily and you say it’s trash. I couldn’t find an email from you using it since 2013. I guess we figured something out. NNs are cool too; at last count we use half a dozen different ones including GNNs… NeSy is hot and I can hardly read a paper these days that doesn’t talk about triples.
Re:MillenniumDB: A graph database engine using domain graphs on Friday January 31, 2025 @06:39PM (#42893804)
OWL ontologies making a big comeback as part of Knowledge Graph groundings for LLM outputs. And several SPARQL and RDF knowledge graph startups are VC-baked and thriving. The world is a big place.

Older »