Stories by pk-protect-ai

From 300KB to 69KB per Token: How LLM Architectures Solve the KV Cache Problem

Show HN: Unwind – I built a security proxy for AI agents on a Raspberry Pi

RegalMail – give your AI agents a real email inbox

Idea Killshot: stress-test your startup idea before the market does

Show HN: Mengram – Open-source memory layer for AI agents

GrantAi – Deterministic memory for AI agents (exact recall, 90% fewer tokens)

Ask HN: Should training bottleneck detection be a product or just a feature?

Show HN: Vex Runtime reliability layer that auto-corrects AI-Agent hallucination

AI Has Amnesia. You're Paying. Blame the Architecture

Are there any companies who are anti-AI?

Nano Banana 2

Show HN: A framework to observe epistemic drift in AI outputs

My AI kept lying to me, so I built a stress test for agents

Show HN: Mengram – AI agent memory with facts, events, and evolving workflows

Show HN: OpenClaw remembers for OpenClaw. Sekha remembers for your full workflow

Show HN: Built an AI tool that routes tasks to agents, humans. Am I crazy?

No one has defined the "AI Strategist" role – so I wrote an OSS book to do it

Palantir's secret weapon isn't AI – it's Ontology. An open-source deep dive

An methodology for new business development in the GenAI era

Ask HN: Why does single-node DDP sometimes get slower with more GPUs?

1 2 3 4 5 6 7 8 9 10