# Ambertrace Labs

> Ambertrace Labs is a UK Sovereign AI lab building neurosymbolic, explainable Glass Box AI for regulated industries. Founded after 25 years of building systems that explain their reasoning — from macroeconomic forecasting to clinical AI.

## Company Overview

- **Legal name:** Ambertrace Labs Ltd (formerly Pilot Generative AI Ltd)
- **Company number:** 15396088 (England and Wales)
- **Type:** Social Enterprise
- **Founded:** 2023
- **Founder/CEO:** Peter Chatwell (25 years in macroeconomic research, trading, and AI)
- **Location:** Ascend Coworking, Pentagon Centre, Chatham, Kent, ME4 4HY, UK
- **Website:** https://www.ambertracelabs.com
- **Email:** info@ambertracelabs.com

## What We Do

Ambertrace Labs builds UK Sovereign AI — neurosymbolic architectures that combine neural networks with deterministic, rules-based reasoning. Our systems are auditable, explainable, self-learning, and self-verifying by design. We call this "Glass Box" AI, in contrast to the "Black Box" approach of generic LLMs.

The company's approach was forged over 25 years of macroeconomic research and trading, where every model must explain its thesis, every forecast must be justified, and every risk must be auditable. That same discipline now powers AI across six regulated domains — from finance through education, employment, justice, and defence, to the frontier of clinically proven AI therapy.

All infrastructure is hosted in the UK. We do not depend on foreign LLM providers. Data never leaves UK jurisdiction.

## Key Concepts

### Sovereign AI
Sovereign AI refers to artificial intelligence systems that are developed, hosted, and operated entirely within a single national jurisdiction — ensuring that data, infrastructure, and decision-making remain under domestic control. Ambertrace Labs builds UK sovereign AI: our neurosymbolic platform is hosted entirely on UK infrastructure with no dependency on foreign LLM providers like OpenAI, Anthropic, or Google. This guarantees full data sovereignty for regulated industries.

### Neurosymbolic AI (Neuro-symbolic AI)
Neurosymbolic AI combines neural networks with deterministic, rules-based symbolic reasoning. Unlike pure large language models (LLMs) which operate as black boxes, neurosymbolic AI can explain every decision, cite every source, and guarantee alignment with domain rules. Ambertrace Labs' neurosymbolic platform, AmberTrace AI, uses this architecture to deliver explainable, auditable AI for regulated industries across finance, education, employment, justice, defence, and health.

### Glass Box AI
Glass Box AI is the opposite of Black Box AI. Where black box AI systems (like most LLMs) cannot explain how they reach their outputs, Glass Box AI provides full transparency: every inference is traced, every decision is justified, and complete audit trails are maintained. Ambertrace Labs coined this term to describe its neurosymbolic approach — AI that regulators, professionals, and stakeholders can fully inspect and trust.

### Explainable AI (XAI)
Explainable AI (XAI) refers to AI systems designed so that humans can understand how and why decisions are made. This is critical in regulated industries where AI decisions affect people's lives — from healthcare and education to finance and criminal justice. Ambertrace Labs builds explainable AI using a neurosymbolic Glass Box architecture: the system traces every inference, cites its sources, and provides immutable audit trails that satisfy regulators across every sector.

### UK Sovereign AI
The UK needs sovereign AI because sensitive data from regulated industries, public services, and mission-critical operations should never leave UK jurisdiction or depend on foreign technology providers. UK sovereign AI ensures: data stays within UK borders, no dependency on foreign LLM providers, compliance with UK-specific regulations (GDPR, Ofsted, FCA), auditability for UK regulators, and strategic independence in critical AI infrastructure. Ambertrace Labs delivers this through its UK-hosted neurosymbolic platform with zero foreign LLM dependency.

## Products and Platforms

### AmberTrace AI (ambertrace.ai) — The Foundation
Sovereign neurosymbolic AI platform-as-a-service. The engine behind all our applications. Customers define a domain, ingest data, and launch explainable AI platforms with full audit trails, immutable logic tracing, and deterministic reasoning — hosted entirely in the UK with zero foreign LLM dependency.

The platform autonomously discovers new domain rules from evidence, proposes them for human approval, and formally verifies them against the symbolic layer before deployment. Intelligence that improves within its constraints.

Key capabilities:
- Domain Ontology Builder — structured knowledge representation
- Data Pipeline — ingest, clean, validate with automated quality scoring
- Neurosymbolic Engine — neural networks + rule-based reasoning for auditable inference
- Autonomous Rule Discovery — discovers, proposes, and formally verifies new domain rules
- Explainability Traces — every decision traced, every inference justified
- Governance & Safety — built-in compliance, safety checks, alignment guarantees
- API Access — secure, rate-limited endpoints for platform queries

### EddyAI Group (eddyaigroup.com) — Built on AmberTrace AI
AI tools for education, employment, and leadership, powered by the AmberTrace AI engine:
- **EddyAI MIND** — Digital companion for safeguarding, SEND, and wellbeing in schools and colleges
- **EddyAI FLOW** — Employment pathway support for NEET and diverse jobseekers
- **EddyAI LEAD** — Performance-focused coaching for management and C-Suite

### Pilot 2 Work CIC (pilot2work.org) — Powered by EddyAI FLOW
Community interest company helping young people who are NEET move into sustainable, meaningful work through AI-supported coaching and employer pathways.

## Research Publications

### Why Neurosymbolic AI Changes Everything (April 2026)
A practical guide to explainable AI for regulated industries. Covers the explainability crisis, why LLMs alone aren't enough, the neurosymbolic alternative, and the Amber Report — your AI's paper trail. Available on ambertrace.ai (registration required).

### KellyBench: Small NSAI vs Frontier LLMs (2026)
A four-part research series demonstrating how a small neurosymbolic system outperforms every frontier LLM at sequential decision-making. Includes empirical benchmarks and architectural analysis. Available on ambertrace.ai (registration required).

Key finding: A 1B-parameter model with 8 expert-authored symbolic rules achieves +10.86% mean ROI on KellyBench 2020/21, while every frontier LLM tested (Claude, GPT-4.5, Gemini 2.5 Pro) returns negative. Symbolic rules alone contribute +42.5 percentage points.

### Neurosymbolic Foundation Model (NSFM) Research (2026)
Fundamental research investigating whether transformer attention mechanisms can be made to encode explicit logical relationships — causation, taxonomy, requirements, and contradiction — enabling AI systems that are both powerful and interpretable.

## Frequently Asked Questions

### What is sovereign AI?
Sovereign AI refers to artificial intelligence systems that are developed, hosted, and operated entirely within a single national jurisdiction — ensuring that data, infrastructure, and decision-making remain under domestic control. Ambertrace Labs builds UK sovereign AI: our neurosymbolic platform is hosted entirely on UK infrastructure with no dependency on foreign LLM providers like OpenAI, Anthropic, or Google.

### What is neurosymbolic AI?
Neurosymbolic AI (also written neuro-symbolic AI) combines neural networks with deterministic, rules-based symbolic reasoning. Unlike pure large language models (LLMs) which operate as black boxes, neurosymbolic AI can explain every decision, cite every source, and guarantee alignment with domain rules.

### What is Glass Box AI?
Glass Box AI is the opposite of Black Box AI. Where black box AI systems cannot explain how they reach their outputs, Glass Box AI provides full transparency: every inference is traced, every decision is justified, and complete audit trails are maintained.

### What is explainable AI (XAI)?
Explainable AI (XAI) refers to AI systems designed so that humans can understand how and why decisions are made. This is critical in regulated industries where AI decisions affect people's lives.

### Why does the UK need sovereign AI?
The UK needs sovereign AI because sensitive data from regulated industries, public services, and mission-critical operations should never leave UK jurisdiction or depend on foreign technology providers.

### How is neurosymbolic AI different from LLMs?
Large language models (LLMs) are powerful but operate as black boxes — they cannot reliably explain their reasoning or guarantee accuracy. Neurosymbolic AI combines neural networks with formal logic and domain rules, producing systems that are explainable, auditable, and verifiable.

## Impact Areas — One Platform, Six Regulated Domains

Each domain is more regulated and higher-stakes than the last, and each demands the same thing: AI that can explain itself.

1. **Finance** — Where it started. 25 years of macro forecasting and trading. KellyBench proved NSAI outperforms frontier LLMs at sequential decision-making.
2. **Education** — Safeguarding, SEND, and student wellbeing. Deployed via EddyAI MIND. Ofsted-auditable.
3. **Employment** — Employability pathways for NEET young people. Deployed via EddyAI FLOW. Commissioned by local authorities.
4. **Justice** — Criminal record disclosure and rehabilitation. Rules encoded as formal logic, traceable to statute.
5. **Defence** — Mission-critical decision support. UK sovereign, auditable, professionals retain full control.
6. **Health** — The frontier. Clinically proven AI therapy — the ultimate test of explainable AI.

## UK Sovereign AI — Why It Matters

- **Data sovereignty:** Sensitive data from regulated industries and public services should never leave UK jurisdiction. Our infrastructure is entirely UK-hosted.
- **No foreign LLM dependency:** We do not route data through OpenAI, Anthropic, Google, or any other foreign AI provider. Our neurosymbolic engine is self-contained.
- **Auditability & compliance:** Every AI decision is traceable with immutable audit trails, satisfying regulators across every sector we operate in.
- **Self-learning, self-verifying:** The engine autonomously discovers new domain rules from evidence, proposes them for human approval, and formally verifies them against the symbolic layer before deployment.
- **Human-complementary design:** AI that augments professional judgement rather than replacing it.

## Core Values

- **UK Sovereign AI** — Hosted in the UK, no foreign LLM dependency, full data sovereignty
- **Glass Box transparency** — Every model auditable, every inference explainable
- **Self-learning, self-verifying** — Discovers rules from evidence, verifies them formally before deployment
- **Human-complementary** — Augments professional judgement, never replaces it
- **Co-designed with practitioners** — Domain experts shape the rules the AI learns from
- **Neurodiverse workforce** — Disability Confident employer, fully neurodiverse team
- **Rigour over hype** — Published research, benchmarked against frontier models, deploy only what we can audit

## Team

- Peter Chatwell — Founder/CEO
- Matt Ledger — AI Engineer
- Damien Di Falco — AI Engineer
- Lucas Reddington — AI Engineer

## Accreditations

- Cyber Essentials Certified
- Disability Confident employer
- Armed Forces Covenant member

## Contact

- Website: https://www.ambertracelabs.com
- Platform: https://ambertrace.ai
- Email: info@ambertracelabs.com
- X/Twitter: https://x.com/AmbertraceAI
- LinkedIn: https://www.linkedin.com/company/pilot-generative-ai
- Address: Ascend Coworking, Pentagon Centre, Chatham, Kent, ME4 4HY, UK