PSNF-Core

A transparent, browser-only cognitive AI.
No cloud. No black box. You control what it learns.

Client-side (runs in your browser) 🔍 Inspectable (concepts, links, facts) ✍️ User-controlled learning

Not an LLM. Not a chatbot trick.

PSNF-Core is a cognitive engine that works on concepts, relations and roles. It can feel “non-deterministic” because its internal state evolves over time (memory, salience, context), but its reasoning is observable and debuggable.

Tip: teach it explicitly (Store as Fact) · explore links · inspect memories

What you can do today

  • Ask “what/why/how” questions
  • Store user facts (controlled learning)
  • Search concepts & semantic links
  • Inspect STM, episodic memory, roles

Transparent graph · Semantic spine · Memory tools

🧠 Cognitive, not generative

PSNF-Core does not “autocomplete” the world. It builds answers by navigating a concept graph, semantic roles (agent/action/patient) and memory.

🔍 Fully transparent

Inspect concepts, links, facts and memory traces. No hidden weights, no opaque neural layers.

Concept graph · Roles · Episodic memory · Debug view

✍️ User-controlled learning

You decide what becomes knowledge. Store facts explicitly. Reset whenever you want. (Free edition can keep memory simple by design.)

🧩 Semantic search

Explore a concept like dog: outgoing links, incoming links, and stored user facts.

🛡️ Privacy by architecture

Runs locally in the browser. No server required. Ideal for education and experimentation.

🚀 Built for evolution

Roadmap includes richer rules (part-of, causes, used-for), summaries, concept maps, personality control, multilingual input.

Who is PSNF-Core for?

Researchers · Educators · Developers · Curious minds. If you want to understand how an AI reasons—not just what it outputs—this is for you.

Transparent cognition > black-box generation

New here?

Start with the explanation page, then try the demo and store a fact:

“A dog is an animal.” → Store as Fact → “What is a dog?”