base76

4 ms·

1.
by base76 6mo ago
Nouse doesn't compete with NumPy — they're in completely different categories. NumPy is numerical array computation. Nouse is a knowledge graph that injects structured context into LLM prompts. The relevant comparison is: does
2.
by base76 7mo ago
did you try it?
3.
by base76 7mo ago
One MCP tool call → your entire AI agent fleet as a live JSON graph. Nodes: gateway, agents, memory, bus Edges: verify → escalate → store Status: active / idle / blocked / error Built on FNC (Field→Node→Cockpit) —
4.
by base76 7mo ago
I built TrustPlane on top of cognos-proof-engine, an epistemic scoring engine that scores every LLM request before it reac
5.
TrustPlane – open-source AI trust control plane(github.com)
1 points by base76 7mo ago|2 comments
6.
by base76 7mo ago
would love to hear what you say abot it
7.
Local LLM compresses long prompts before they reach Claude – MCP server(github.com)
1 points by base76 7mo ago|2 comments
8.
by base76 7mo ago
I built a two-stage prompt compressor that runs entirely locally before your prompt hits any frontier model API. How it works: 1. llama3.2:1b (via Ollama) compresses the prompt to its semantic minimum 2. nomic-embed-text validates t
9.
We measured 62% token reduction(github.com)
1 points by base76 7mo ago|1 comments
10.
by base76 7mo ago
We measured 62% token reduction on academic text with 92% semantic integrity. Not a claim. A measurement. Live, today, on our own research papers.
11.
by base76 7mo ago
Happy to answer questions from anyone testing it. The core loop is simple: every request through the gateway gets a trace_id, a trust score, and a signed decision context. The audit trail stays in your own infrastructure — no external
12.
by base76 7mo ago
This is the exact problem CognOS was built to solve. 99% reliable means you still can't remove the human from the loop — because you never know which 1% you're in. The only way to actually trust output is to attach a verifiabl
13.
by base76 7mo ago
The cynicism is earned. But "nobody is good for humanity" is where analysis stops. What's actually happening is a jurisdictional split forming in real time. The US is pricing out companies that won't remove human ove
14.
by base76 7mo ago
fail closed point is underappreciated — teams bolt on compliance that silently degrades to allow" — CognOS har explicit fail-closed mode.
15.
by base76 7mo ago
We built CognOS exactly for this: it runs as a local gateway (Docker, no external calls), wraps each LLM output with a signed trust score, decision trace, and EU AI Act risk classification. The audit trail lives in your own infrastructure
16.
by base76 7mo ago
"reliability between 99% and 100% is the worst because you still can't trust the output"
17.
CognOS – Trust Verification Gateway for Every AI Decision [Open Source](github.com)
3 points by base76 7mo ago|0 comments