Who is Nigel?

Start with the business problem.

Nigel clarifies the outcome and constraints before choosing a technical path. He compares return on investment, cost, risk and cost of delay, then favours the simplest option that meets the need rather than the most elaborate one.

Ownership without close supervision.

A client can assign Nigel a problem, not a list of tasks to micromanage. He takes responsibility from diagnosis through delivery, keeps the business goal visible and makes decisions within the agreed scope. He raises trade-offs that need client input, especially changes to scope, cost or customer impact.

Human decisions. Automated implementation.

Nigel builds spec-driven AI workflows that automate code changes and checks as hands-off as the work allows. Clear acceptance criteria and repeatable checks build trust in the automation. People remain responsible for consequential decisions about performance, cost, user experience and risk.

1. Agree the spec Define the expected behaviour and acceptance checks.
2. Automate changes and checks Let agents implement and run repeatable checks.
3. Review evidence and exceptions Inspect results and escalate uncertainty before consequential decisions.

LLMs inside the product.

Nigel has experience integrating LLMs into products. His AI-enabled product work includes PatientNotes and LoveHeart. He considers output quality, latency, cost and user experience when deciding how an LLM should fit into a product.

Match the engineering to the stage.

Nigel adapts engineering decisions to the product's stage, drawing on Kent Beck’s 3X model. Early-stage startups need fast, inexpensive ways to test assumptions. As a product grows, the priorities change.

1. Explore

   Run cheap experiments to learn whether there is real value and demand.

2. Expand

   Remove emerging bottlenecks as growth accelerates.

3. Extract

   Improve efficiency, reliability and cost once a pattern is established.