ZPL – a deterministic engine for binary-matrix bias scoring

cicicalex1 pts2 comments

Zero Point Logic | Stability scoring for any system<br>Skip to main content

⓪ Live Engine<br>Watch CoreBits converge to the attractor.<br>Every "+" is a metabinary input drawn toward equilibrium by deterministic gravity. Same engine running in production — not a mockup.

p→0.5ATTRACTOR<br>inputs: 0<br>new/s: 0<br>consumed/s: 0

unstablewarningstable<br>cycle ↻ 18s · 140 CoreBits · live

> ZPL ENGINE<br>COREBIT v4.2 · stream

✅ Deterministic — same input, same orbit✅ Live — running now in your browser✅ Not quantum — classical metabinary✅ Attractor p=0.5 — proven across 2.6B trials

What AIN tells you<br>AIN Score<br>0–100<br>How stable is your data right now? A single number from 0 (collapse) to 100 (perfectly balanced). No dashboards, no walls of metrics — one answer.

Status flags<br>STABLEInside the neutral attractor — predictable, low drift.

ACTIVEDrifting away from balance — worth watching, possibly trending.

INHIBITEDCollapsing — bias is concentrated, system is breaking.

What it answers<br>ZPL was built to answer one question: is this system about to break? It returns a deterministic, mathematically grounded answer in milliseconds — not a hunch, not a black-box ML score.

Use casesOne engine, every domain.<br>AIN doesn't care what the bits represent. Same engine, same call, six very different problems.

Crypto Markets<br>Detect when a market is about to flip. Live AIN overlays on every major coin — see it running on finance.zeropointlogic.io.<br>Learn more

Game Economies<br>Stop hyperinflation in MMOs and web3 games before it crashes your TVL. Score loot tables, drop rates, matchmaking fairness.<br>Learn more

AI / ML Models<br>Audit model bias and drift in production. Score any sequence of outputs — LLM responses, classifier decisions, agent loops.<br>Learn more

Security<br>Detect anomalies in network traffic, auth logs, or access patterns. Stability collapse is often the first signal of a breach.<br>Learn more

Research<br>Analyze convergence and consensus in sequential experimental data. Open methodology, deterministic outputs — reproducible by design.<br>Learn more

Forex / Macro<br>Monitor stability of currency pairs, sovereign bonds, and central-bank rate paths. AIN flags regime shifts before the press conference.<br>Learn more

How it worksThree steps. One number.

Send your data<br>POST your normalized binary input to /compute. Official SDKs in TypeScript and Python — any other language hits the REST API directly.

Engine computes AIN<br>Sub-50ms response with ain (0–1), ain_status, and engine status. Deterministic — same input, same output, every time.

Act on the signal<br>Plug the result into your monitor, alerting, trade flow, or training loop. No dashboards required.

zpl.ts<br>Copy<br>// 1 — call the engine<br>const result = await zpl.compute({<br>matrix: yourMatrix, // N×N of 0s and 1s, 3 ≤ N ≤ 100<br>samples: 1000,<br>});

// 2 — read the score<br>console.log(result.ain); // 0.0..1.0 (higher = more neutral)<br>console.log(result.biasLevel); // none | low | moderate | high | critical<br>console.log(result.status); // STABLE | ACTIVE | INHIBITED

// 3 — act on it<br>if (result.ain 0.35) alertOnInstability();<br>Sample SDK call. Official SDKs ship TS + Python; any other language calls the REST API directly.

Use it everywhereOne engine. Four ways to call it.<br>Web dashboard for clicks. MCP for Claude. CLI for shell pipelines. SDK for TypeScript / Python. All hit the same Rust engine — same scores, same plan limits, same trade-secret formula on the server.

WEB<br>Dashboard<br>Browser-only. Sign in, paste a matrix, get an AIN score. No install.<br>open in browserRead more →MCP · v4.1.9<br>MCP Server<br>Plug into Claude Desktop, Cursor, Windsurf. 68 tools — Claude picks one automatically.<br>npx zpl-engine-mcp setupRead more →CLI · v1.1.9<br>Terminal CLI<br>One-shot scoring from shell, scripts, CI. Pipe stdin, get JSON. 20 commands.<br>npx -y zpl-engine-cliRead more →SDK · v2.0.4<br>TypeScript & Python<br>Typed clients with auto-retry, batch helpers, client-side validation. npm + PyPI.<br>npm install @zeropointlogic/sdkRead more →<br>Or call the REST API directly — OpenAPI 3.0 spec generates clients for any language.

PricingStart free. Scale when ready.<br>Pay only for compute tokens. No seat fees, no hidden costs. Cancel any time.

Free<br>$0forever<br>5,000 tokens / month. One API key. Public endpoints. Perfect for prototyping.<br>Get started<br>Most popular<br>Pro<br>$29/ month<br>50K tokens. 3 API keys. gRPC streaming. The default for production work.<br>Start Pro<br>Studio<br>$149/ month<br>500K tokens. 10 keys. Higher dimension limits — built for game economies and ML pipelines.<br>Start Studio<br>Enterprise<br>$499/ month<br>10M tokens. SLA, dedicated support, custom contracts. Talk to us.<br>Contact sales

Compare all 8 plans

Why ZPLBuilt on math, not vibes.

Open science<br>The AIN framework is published with a Zenodo DOI. The why is public, the how is ours.<br>Read the methodology<br>Self-hosted<br>Runs on Hetzner bare metal. No third-party SaaS chain, no surprise outages from a vendor you've never heard of.

Deterministic<br>Same input always returns the same output. No model drift, no random...

engine deterministic score learn result input

Related Articles