Selected builds

Ideas tested by making them real.

I use small products to examine practical questions about AI: where it helps, where the surrounding experience matters, and what evidence is needed before trusting the result. The case studies show the decisions, limitations, and next tests behind the strongest work.

Try the launch review

AI Product Launch Review

Turn an AI idea into a launch decision you can test

Live

Next.jsTypeScriptDecision rulesEvaluation design

Try the launch review →Read case study →

What it is

An interactive product review that turns a system description into explicit claims, boundary tests, evidence requirements, and launch gates. The output is designed to be challenged, checked, and exported.

Why I built it

AI product reviews often begin with broad promises and end with polished documents. I wanted to move the hard part earlier: define what must be true, how it will be tested, and who needs to see the evidence before launch.

What it taught me

A useful review should make claims falsifiable. The product becomes more trustworthy when its decision rules and limitations are visible, and when the output names evidence rather than manufacturing confidence.

ai-governance-agent-xi.vercel.app

AI Governance Assessment Agent

Automated first pass AI risk and governance reviews

Live

Next.jsAnthropic APIVercel

ai-governance-agent-xi.vercel.app ↗Read case study →

What it is

An AI tool that reviews supplied system documentation against a selected governance framework and drafts a structured assessment. It surfaces risks, missing information, and questions worth following up.

Why I built it

I wanted to test whether AI could make a document-heavy governance review easier to begin. The challenge is making its reasoning, sources, and gaps visible enough for a person to check.

What it taught me

A tidy assessment can look more certain than the evidence behind it. Clear questions, traceable inputs, and room to say that information is missing matter as much as the generated answer.

scenariolab.quest

Fable

An AI communication coach for hard conversations

Live

Next.jsTypeScriptAnthropic APIVercel

scenariolab.quest ↗Read case study →

What it is

An AI communication coach for practising difficult conversations. Work through scenarios, try a different approach, and get feedback on your delivery.

Why I built it

I wanted a place to rehearse a conversation before having it for real. Building Fable let me explore how personas, prompts, and feedback can make an AI interaction useful beyond an open chat box.

What it taught me

The surrounding experience matters: the scenario, the instructions, and the way feedback is presented. A model-generated score is a prompt for reflection; its usefulness still needs to be judged against the conversation you actually want to have.

A calmer start to the morning

Northbound Notes

A morning briefing that writes itself

Live

Claude CodeScheduled taskWeb searchGmail

Read case study →

What it is

A daily email I use to start the morning: Toronto weather, news, developments in AI, and a small selection of stories worth a closer look. Built with AI to bring some structure to a noisy information feed.

Why I built it

I was piecing together the morning from several different sources. I wanted one readable starting point, with enough context to decide what deserves more attention. It is the project I use every day.

What it taught me

Editing is the product. Pulling the news in was the easy half. The hard half was deciding what to cut so the whole thing still reads in two minutes, and writing it so it sounds like a person rather than a feed.

GRC Risk Register Tool

A lightweight risk register that drafts itself

In progress

PythonAnthropic API

What it is

An experiment that turns a plain-language risk description into a draft register entry, with fields for likelihood, impact, controls, and ownership.

Why I built it

I wanted to explore how a small tool could make risk documentation easier to begin. The draft gives someone a structure to question and refine; it cannot establish the risk on its own.

What it taught me

Constraints beat cleverness. A tight, opinionated schema produced far more useful output than an open ended 'assess this risk' prompt.

Notion & Productivity Templates

Digital templates sold on Etsy

Live

NotionEtsy

What it is

A small catalogue of Notion templates, including a YouTube workflow hub and productivity systems, packaged and sold as digital downloads.

Why I built it

My first taste of shipping something people actually pay for. Low stakes, real feedback, and a crash course in positioning and digital product distribution.

What it taught me

Distribution is a skill of its own. Making the thing is maybe half the work. Getting it in front of the right person is the other half.