We were good at this before AI. We're better at it now.

Small, senior teams of product designers and engineers who know where AI accelerates the work and where it doesn't.

The work has changed. The bar hasn't.

Coding agents produce competent output in most languages and frameworks. Codebases that took weeks to analyse take hours. Prototypes that took a sprint take an afternoon.

The things that were hard to commoditise before are more valuable now. Framework knowledge is cheap. Engineering judgement, design thinking, and consulting instinct aren't.

How the team works

An engagement brings together senior engineers and product designers in a tighter loop than either worked in a year ago. Design decisions feed specifications that coding agents act on. Engineers push back on product direction with prototypes built in hours rather than sprints. Hand-offs that used to take days now happen in conversations that take minutes.

That only works with the right people in it. Seniority on both sides. Designers who understand implementation trade-offs. Engineers who take product thinking seriously. Generalists rather than specialists.

It also works with fewer people. An engagement that used to staff five now staffs three. Each person covers more ground because AI absorbs the work that used to need another body, and seniority absorbs the judgement AI can't supply.

What's different about an engagement today

Three specific things have changed compared to two years ago.

Senior teams, smaller shape

• An engagement that used to staff five now staffs three.
• More senior people, covering more ground. AI absorbs the work that used to need another body. Seniority absorbs the judgement AI can't supply.
• A shape-of-team call, not a cost call.

Discovery got more valuable, not less

• Cheap code raises the cost of building the wrong thing.
• Discovery, design, and clear specification keep an AI-accelerated team aimed at the right problem.
• Specifications now do double duty: inputs for AI code generation as well as documents for humans.

Workflow, not tools

• A team with agentic AI licences and no workflow discipline is just a team with agentic AI licences.
• The gain comes from redesigning the work. How problems get decomposed. How specs get written. How AI outputs get reviewed. How tests get structured to catch what AI gets subtly wrong.
• Our teams work this way. That's the difference.

Where we don't use AI

Knowing when not to use AI is as important as knowing when to. Here's where we draw the line.

Architecture

Decisions with long-term consequences are made by people, not generated. We don't let AI set the direction on things the client will be living with in three years.

Security

Critical paths get close engineering review, not automated generation. Code that handles authentication, payments, or sensitive data gets read, understood, and signed off by a senior engineer.

Client conversations

Conversations that shape an engagement stay with our people. The judgement that comes from understanding a client's context, constraints, and history doesn't get delegated to a tool.

Inherited complexity

What AI introduces gets surfaced and understood, never buried in code someone else will inherit. If a model produces something we can't fully explain, we don't ship it.

Same People, New Model

If you've worked with us before, you'll recognise the engagement. We've changed how we build. We haven't changed what we stand for.

What's the Same

• The standard for delivery quality
• The discovery-led approach
• The commitment to being the team that stays useful across a full engagement, not just the first six weeks.

What’s New?

• Smaller Teams
• AI in the workflow
• Faster prototyping
• Sharper specifications

Our Work

Web3

Scaling security and reliability with DevOps maturity

YLD delivered strategic recommendations to enhance security and reliability at scale that would enable faster innovation and better risk management.

Learn More

Research

Accelerating scientific progress through digital innovation

ASAP and YLD created a secure digital platform for collaborative research, establishing a technical benchmark and agile workflow to support the scientific community.

Learn More