AI Engineering
Dizputed — Co-Founder
Building Legal Technology from the Ground Up
In 2024 I co-founded Dizputed, a legal technology platform that helps separated families manage and analyse communications in family law proceedings.
Starting from scratch, I retrained in Python, software architecture, LLM deployment and product development — building the application end-to-end alongside my co-founder.
The platform processes thousands of messages, classifies conduct patterns using multi-stage AI pipelines, and generates structured legal reports used in court proceedings.
Dizputed — Creative Marketing
The Dizputed ABC Book
Not every product story needs a demo video. For Dizputed, I created an illustrated ABC Book — a deliberately playful, character-driven way to communicate a serious product to a stressed audience.
Each letter tells a story about the product's features through accessible, human language. A is for Analysis. B is for Behaviour. C is for Court.
The book demonstrates that strong creative thinking and technical product development are not in opposition — they reinforce each other.
AI-Powered Analysis
Multi-Stage Message Intelligence
The core of Dizputed is a multi-stage AI pipeline that imports, classifies and analyses thousands of messages from sources including WhatsApp, Gmail, SMS and PDF.
Each message is processed through a series of agentic LLM steps: conflict detection, tone classification, pattern identification and conduct scoring.
Designing these agentic workflows — from prompt engineering to output validation — sits at the intersection of software architecture and applied AI.
Automated Reporting
LLM-Generated Legal Reports
Dizputed's report engine takes structured AI analysis and transforms it into professional Word documents ready for use in family law proceedings.
The system draws on a database of classified messages, applies LLM summarisation and narrative generation, and assembles the output into a structured, court-ready format.
Building this required mastering the full stack: database schema design, batch processing, prompt architecture, document generation and UX.
Architecture & Agentic Workflows
From Creative Brief to Technical Architecture
My approach to software mirrors my approach to strategy: start with the problem, not the technology. I design systems that are legible, maintainable and built to evolve.
This includes agentic AI workflows — chains of LLM reasoning steps with structured outputs, validation loops and human-in-the-loop checkpoints where precision matters.
I also advise growth-stage companies as a Fractional CMO on the commercial and strategic application of AI — from go-to-market positioning to building AI-native content operations.
