Expertise
My work today centres on secure, production-ready agent systems at Nearfield.ai, built on a foundation of declarative languages and compiler technology developed at Aucoda over the past decade.
Nearfield.ai — Enterprise Agent Systems
PIIQ — Enterprise Compliance
A desktop application that automates Data Subject Access Requests (DSARs) and FOIA workflows, reducing a multi-day manual process to hours.
PIIQ uses locally deployed AI models for PII discovery and redaction, so sensitive data never leaves the organisation. It is a demanding domain — requiring precision, traceability, and strict data handling — and proves that agent systems can operate reliably in production.
OS/A — Operating System for Agency
A proprietary multi-agent infrastructure that supports reliable, structured execution of intelligent agents.
OS/A provides the runtime and orchestration layer beneath Nearfield's enterprise products, designed for agent systems that must operate securely on local infrastructure.
The Contextualiser
Episodic machine memory for agent systems — giving agents durable, structured recall of what has happened.
This builds directly on my long-running research interest in machine memory and capturing context, applied to keeping multi-agent execution coherent over time.
Agent Markdown
A structured format for authoring and constraining agent behaviour.
Agent Markdown is part of the infrastructure Nearfield is preparing to open-source as it transitions toward a broader developer platform.
Scope
A new class of development environment designed to safely manage, structure, and constrain code generated by intelligent agents in production.
Scope guides and constrains the output of autonomous agents so that agent-generated code can be trusted in real engineering workflows.
Aucoda — The Au Language Ecosystem
Au Programming Language
A declarative language where code is written once and automatically transformed into multiple target languages, including C++, Rust, Swift, Java, JavaScript, and Python.
Au is an example of Automatic Coding done right — it takes every other language and uses it as its instruction set. I put graduates and undergraduates, in their first jobs, to work writing complex software for clients including the NHS, PwC, HMRC, the Metropolitan Police, and Marks & Spencer.
Outputs are produced via the companion language Autron (Automatic Transformation), which can also emit Solidity and other targets. Using Autron you can teach Au a new language in about two weeks.
Autron
A companion language to Au that specialises the output of the compiler to fit any purpose.
Autron writes out other languages — Python, Java, JavaScript, C++, C#, HTML, CSS, SQL, CouchDB, Swift — and lets you implement any language not on the list. You can also use it to analyse and optimise code, improve standards, or enforce security checks.
Automic IDE
A fully browser-based IDE for Au — create projects, run, debug, and "forge" code into the target output, with Git integration for full source control.
Au Lab lets you test blocks of code — data processing, rules, external API calls — before inserting them into the main codebase, without the distraction of navigating the full application.
Au Lab Notes is a Markdown editor that mixes text and Au together, for research books, documented experiments, or self-contained tutorials (every tutorial in the Introduction to Programming with Au course is a Lab Note).
Au Lambda — AWS Integration
Au runs as a Lambda function on AWS, so you can compile from anywhere with no infrastructure — wherever you see Au, you can run it, debug it, change it, and run it again.
Knowledge and Learning Engine (KLE)
The platform behind "Introduction to Programming with Au" — a 100k-word course with code, text, quizzes, exercises, skills builders, and progress monitoring, running in the browser.
I use the KLE to build bespoke training courses for clients — training a student in Au and then in a client's domain problem, so the client can employ that developer with confidence.
Au Active is a set of 25 projects developers use to gain experience, each with a full professional specification, wireframes, and UI design — getting students working as they would in industry rather than coding maths puzzles.
xCAT Cryptocurrency
As part of the planned rollout of Au I designed a cryptocurrency called xCAT — a DAO to give underprivileged young people access to equipment and broadband so they can learn Au. xCAT was not released given prevailing financial conditions, but the intent remains to level up under-served communities around the world.
Catalyst Job Portal
Catalyst is a job portal where Au developers can showcase their journey and their Au Active projects — bringing people with the skills together with the people who need them.