Hyperdimensional Computing Engine

Encode meaning
into vectors

Transform text into 10,000-dimensional binary hypervectors using brain-inspired computing. Deterministic, stateless, and built for production workloads.

Try the playgroundAPI Reference

Built for production

A compute-only microservice with no persistent state. Deploy anywhere, scale horizontally.

10,000-D Vectors

Encode any text into a 10,000-dimensional binary hypervector for rich semantic representation.

Stateless Compute

No database, no sessions. Pure deterministic compute. Every request is independent and horizontally scalable.

Secure by Design

API keys live server-side only. All requests are proxied with strict input validation and rate limiting.

Cosine Similarity

Compare any two hypervectors instantly. Measure semantic distance between texts with a single call.

How encoding works

See how AXIOM transforms text into binary hypervectors step by step. Select a pair, then click through each stage of the encoding pipeline.

How Hyperdimensional Encoding Works

Move your mouse over the sphere. Click Play to step through the pipeline.

Compare:
A: "cat"
B: "cat"
Encoding Pipeline
1
Tokenize
Split text into characters
2
Map Atoms
Each char → random binary vector
3
Permute
Rotate each atom by its position
4
XOR Bind
Combine all into one vector
5
Compare
Hamming similarity between vectors

Simple API

encode.ts
const res = await fetch("/api/encode", {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({ text: "quantum computing" }),
});

const { vector, dimensions } = await res.json();
// dimensions: 10048
// vector: Uint8[1256] — deterministic HDC encoding
POST
/api/encode

Encode a text string into a 10,048-dimensional hypervector. Returns a deterministic byte array.

POST
/api/batch-encode

Encode up to 50 texts in a single request. Returns an array of vectors with the original texts.

POST
/api/similarity

Compute cosine similarity between two hypervectors. Returns a float between 0.0 and 1.0.