Co-Founder, Managing Director & Technical Lead · Rheos App Ltd
I built and commercialised Rheos, a multi-tenant system that captures a small business’s context, plans and produces its social content, publishes approved work and tracks how that business appears across social, search and AI answers. It is now used by 50 paying UK businesses. I lead product, web and backend engineering, customer delivery, sales and company operations.
Built and commercialised a multi-tenant AI platform for small businesses. Own the customer-facing product, web and backend systems, AI deployment architecture, customer delivery, sales and company operations.
Analysed FTSE 350 annual reports, governance structures and shareholder outcomes, turning dense public evidence into consistent institutional research.
Coordinated nationwide rail delivery of COVID tests at InterCity Railfreight, then worked as an Accounts Assistant at Annie’s Burger Shack Group.
Research, model evaluation, production software and the work of operating a company.
The progression covered more than software: model evaluation, product design, deployment, sales, customer support, legal work and company operations.
Used ChatGPT from launch for university research, then developed stronger source discipline while analysing FTSE 350 annual reports at ISS STOXX.
Worked across leading OpenAI, Anthropic, Google and open-model families using ChatGPT, Claude Code, Codex and Antigravity; built MCP, session-transfer and multi-model review workflows.
Moved from manually coded websites to a multi-tenant product spanning brand context, planning, generation, approval and publishing. Real businesses were using it by January.
Took the first payment in April and reached 50 paying customers by July. Added authentication, tenancy, billing, observability, recoverable publishing and customer support.
Alongside product and engineering, handled sales, onboarding, support, contracts, trademark, legal work and accountancy, then prepared a Midlands go-to-market rollout.
Rheos App Ltd
A customer starts with their website. Rheos stores editable context about the business—services, audience, tone, proof points, locations, offers and visual identity—then uses it to build a plan and create on-brand copy and visuals.
Customers edit or approve each post before it is published from the same workflow. Analytics inform the next plan, while Search Visibility tracks how the business is represented across social, search and AI answers.
The customer gives Rheos its sources once, adjusts the resulting brand context and works from a repeatable plan rather than a blank prompt. Finished posts stay editable, and nothing represents the business until the customer approves it.
Reports that a weekly content workflow that took hours now takes minutes.
Reports that producing an on-brand post fell from roughly an hour to 10–15 minutes.
My co-founder leads mobile and contributes to technical R&D. I lead product, web and backend engineering, customer delivery, sales and operations. Shared context, specialised agents and connected systems let two founders cover several conventional functions.
Linear, GitHub and a source-aware knowledge system keep plans, code and decisions connected. Codex and Claude support research, implementation, review and QA.
Linear
GitHub
OpenAI Codex
Claude Code
MCP
Multi-tenant application surfaces sit on Firebase and GCP, with asynchronous workers, identity, billing, secret management, observability and explicit release gates.
TypeScript
Next.js
Firebase
Google Cloud
Vercel
Auth0
Stripe
Sentry
General and specialist models cover reasoning, retrieval, image, video, voice and rendering. Providers are chosen by task, cost and quality.
OpenAI
Azure
AWS
fal.ai
ElevenLabs
Perplexity
Firecrawl
Remotion
Prospecting, CRM, booking, support, analytics and customer communication are connected into one operating loop.
Apollo
Gojiberry
Instantly
Attio CRM
Intercom
Cal.com
Amplitude
Resend
Slack
The work that made Rheos a company was the infrastructure around the models: keeping each customer’s data isolated, restricting what agents can do, running long jobs safely in the background, controlling cost, recovering partial failures and keeping a person in control of external actions.
Public examples of my work on persistent context, MCP tools, source quality and human-controlled systems.
Open source · 16 skills
Session transfer, multi-model counsel, MCP development and a delegated search plugin for finding signal in large local AI histories without blowing the context window.
Explore on GitHub ↗
MCP · Investment tooling
Seven read-only tools that give an agent structured access to an investment portfolio while keeping analysis separate from trading authority.
View the MCP ↗
Evidence-first intelligence
A supplement research system designed around provenance, deterministic identifiers, bounded discovery budgets and human review rather than unsupported health copy.
Visit Nutripedia ↗My work at Rheos covers customer discovery, product design, implementation, deployment, incident handling, cost control, support and commercial delivery.