AI work built to survive real workflows.

I help teams move from AI curiosity to working tools, sharper decisions, and repeatable delivery. The examples here show internal transformation work, rapid prototype sprints, and practical automations tested in real use.

What you'll see

Company-wide AI transformation work inside Lightful
Rapid prototype sprints and internal automations
Hands-on side projects built with the same tools I use with teams
Talks and lessons on testing AI ideas in the real world

Approach

Build first. Learn fast.

The work here spans professional delivery, internal transformation, and rapid personal prototyping. Some of it was built inside teams. Some of it was built solo. All of it comes from the same belief: the fastest way to learn what AI is actually good for is to build something usable and put it in context.

I care most about AI when it becomes part of a workflow: removing friction, sharpening judgement, speeding up execution, or opening up a genuinely better way of working.

Professional AI work

From AI interest to embedded practice

These examples show the kind of work I do best: turning broad AI intent into focused experiments, usable tools, and habits that teams can keep using after the first burst of excitement.

Internal AI transformation project screenshot

Lightful, internal transformation

Embedding AI Across the Company

Company-wide AI adoption, built on real friction in real teams.

Measurable time savings across teams

Led a company-wide initiative to embed AI into day-to-day work. The work focused on finding high-friction tasks across teams, then designing experiments, training, and hands-on interventions that helped people adopt AI in practical ways. A key part of the job was enabling teams to build their own tools and workflows instead of waiting on a central bottleneck.

Result: measurable time savings across multiple teams and sustained adoption of AI in day-to-day work.

What this involved

Finding repeatable operational pain points across teams
Running experiments, training sessions, and working interventions
Helping non-specialists build useful AI-supported workflows

What mattered

Grounding AI adoption in real work rather than abstract policy
Balancing speed with enough structure to make adoption stick
Creating momentum that could continue without constant oversight

Lightful, 6-week prototyping sprint

Six Prototypes in Six Weeks

AI product ideas tested in real conditions, fast.

Clear signal on what to build

A focused sprint to test AI product ideas quickly in real-world conditions. Six working AI prototypes were built in six weeks, including tools I built directly and others built with the team. The emphasis was on structured prompts, multi-step workflows, and getting to real signal fast rather than polishing too early.

Result: clear signal on what was worth building, and proof that useful AI products can be created in days, not months.

What we did

Built six functioning prototypes in six weeks
Used each prototype to test a distinct product idea or workflow
Applied AI in ways that were concrete enough to evaluate properly

Why it worked

Compressed learning into days rather than months
Made vague AI opportunities tangible to the wider team
Created a repeatable sprint model for future work

Personal, Slack and calendar automation

Daily Briefings, Delivered Automatically

Calendar in, structured briefing out, posted to Slack.

Less manual planning

A lightweight workflow built to reduce context-switching and improve daily planning. It pulls calendar data automatically, uses AI to generate a structured daily briefing, and posts the output straight into Slack. It is a small example, but it captures the kind of practical internal automation many teams could adopt quickly.

Result: reduced manual planning and a repeatable pattern for simple internal automation.

How it works

Automatically reads the day’s calendar context
Generates a clear briefing rather than raw data
Delivers the result inside an existing communication channel

What it demonstrates

Useful AI automation can be small and sharply defined
Value often comes from structure and timing, not novelty
Simple patterns are easier to spread across teams

Side projects & rapid prototyping

Products that pressure-test the tools

Alongside structured client and team work, I build products to explore ideas and solve real problems. It keeps me hands-on and keeps the tooling, patterns, and workflows honest.

Featured side project

Hailo

Charity, voice AI, multi-agent

Hailo is a calm, voice-led tool that helps people work out what they actually care about, then shows them the charities already doing that work. Instead of a search box that assumes you can name the cause, it runs a short reflective conversation, draws your answers back to you as a single woven thread, and only then suggests organisations that fit.

It is built as three deliberately separate AI agents, so the part you talk to can never recommend or sell, your voice is processed as you speak and discarded, and every match traces back to a line in your own words. A bigger, more ambitious build than the quick utilities below, made solo with AI-assisted tools.

Read the story Try Hailo

The Hailo landing page with the headline find the causes you never knew you cared about.

A woven Hailo circle drawn from the user's answers, labelled your Halo.

Charity match cards showing fit labels and quotes from the user's own conversation.

Charity, evidence AI

Proof Package

Evidence assembly for grant-funded charities, with a printable audit trail.

Private prototype

Evidence assembly tool for grant-funded charities that maps programme documents against funder requirements, flags contradictions and gaps, and produces a printable audit trail.

Built end-to-end using AI-assisted tools, then deliberately kept private because the data-governance questions were too significant for a public prototype.

Sport, mobile

Even Stevens

Grassroots football coaching app that tracks players, substitutions, and playing time. Designed for real-time use on the touchline and built solo using AI-assisted tools.

View project

Maps, utility

Oxford Cycle Parking Map

Simple utility for finding cycle parking in Oxford. Map-based, fast, and mobile-friendly, built to solve a real local problem without unnecessary complexity.

View project

Events, planning

Between Sessions

Conference planning tool that turns a crowded programme into a usable daily plan, maps sessions across Oxford with walking times, and suggests useful stops between events.

View project

The Hinksey Park FC Summer Party tournament app home screen showing live events

Sport, live

HPFC Summer Party

Live tournament app built in a day at a “Dadathon” to run the Hinksey Park FC summer party. Results update instantly, fixtures stay in sync, and parents follow everything live from their phones, across 10 teams and 60+ kids.

“What was different was (Rob and) Jonathan’s amazing tournament app he’s developed that did away with white boards, fixture grids, and maths on the go!” — Tournament coach

View project

How I work

How I pressure-test AI ideas

Start with a real problem

Begin with something people genuinely experience and would notice being solved.

Build something usable quickly

A working prototype creates more learning than a long speculative plan.

Test in real conditions

Put it in front of the people who would actually use it, inside the constraints they really have.

Iterate on what works

Follow the evidence, keep what proves useful, and strip away what does not.

Ways to work together

Bring me in to build, test, and ship.

I work either as a senior product leader inside the team or through focused advisory support around AI, product strategy, and delivery.

Talk about your team

Full-time

Senior product leadership

CPO, VP Product, or Head of Product roles
Leading product, design, data, and engineering work together
Owning strategy, execution, and outcomes rather than just roadmap process
Particularly strong in AI-enabled products and mission-driven organisations

Fractional / advisory

AI & product advisory

Rapid AI opportunity assessment grounded in real workflows
Prototype sprints to test ideas quickly with genuine users
Support for embedding AI into team practice, not just tooling demos
Hands-on guidance where delivery matters as much as strategy

AI work built to survive real workflows.

Build first. Learn fast.

From AI interest to embedded practice

Embedding AI Across the Company

What this involved

What mattered

Six Prototypes in Six Weeks

What we did

Why it worked

Daily Briefings, Delivered Automatically

How it works

What it demonstrates

Products that pressure-test the tools

Hailo

Proof Package

Even Stevens

Oxford Cycle Parking Map

Between Sessions

HPFC Summer Party

How I pressure-test AI ideas

Start with a real problem

Build something usable quickly

Test in real conditions

Iterate on what works

Talks, lessons, and practical walkthroughs

AI-powered discovery

Testing desirability with AI prototypes

Slow down to speed up

Bring me in to build, test, and ship.

Senior product leadership

AI & product advisory