London · Design and engineering since 2014Taking new projects from Q4 2026

Independent digital product studio

We design and build mobile apps, web platforms, AI features and the backend behind them.

A senior London team with twenty years of shipping products for Virgin, Shell, Heycar, Lloyds Banking Group and ClearBank. You work directly with the people doing the work.

Shipped for

Selected products and platforms we’ve helped design, build and ship.

What we do

Four pillars, one team accountable.

Most projects need two or three of these. They are priced and planned together, not handed between agencies. No account manager, no relay of briefs: you speak to the people writing the code.

Services in detail
A mobile product interface being used in a calm architectural space
01Mobile

01Mobile

Cross-platform apps that feel native.

One codebase, both stores, no compromise on gesture, motion or platform convention. Heycar, Virgin Trains Ticketing, Virgin Red and Shell Fleet all shipped on this stack.

What we deliver

  • Product design and prototyping
  • React Native / Expo build
  • Store submission and release management
  • Crash, analytics and OTA update pipeline
React NativeExpoTypeScriptReanimatedEAS
Explore Mobile
A web product displayed on a desktop monitor
02Web

02Web

Platforms that convert and rank.

Server-rendered React with a real design system behind it. Lloyds Mortgages and ClearBank were rebuilt on this approach without a drop in traffic.

What we deliver

  • Design system and component library
  • Next.js / TanStack Start build
  • Core Web Vitals and SEO engineering
  • CMS integration and editor training
ReactNext.jsTanStack StartTypeScriptTailwindDatoCMS
Explore Web
A field of fragments resolving into a precise digital grid
03AI integration

03AI integration

LLM features that survive real users.

Scoped against a real user journey, not a demo. AMiSi.ai runs production generation workflows, while Francis embeds conversational AI inside regulated financial experiences.

What we deliver

  • Feature scoping and model selection
  • Chat, generation and automation flows
  • Evals, guardrails and cost controls
  • Streaming UI and fallback states
AnthropicOpenAIVercel AI SDKpgvectorLangGraph
Explore AI integration
A secure, illuminated data-centre corridor
04Backend

04Backend

The plumbing behind the interface.

APIs, auth, payments and pipelines. Ticketing and banking workloads delivered under FCA and PCI constraints, with runbooks the inheriting team can actually use.

What we deliver

  • API and BFF design
  • Auth, payments and third-party integration
  • Infrastructure as code, CI/CD
  • Observability and on-call runbooks
NodeExpressPostgresSupabaseFirebaseGitHub Actions
Explore Backend
Built with
AI, in production

Features that ship, not demos.

How we scope AI
01Custom workflowsAutomations that run without a person watching, and stop when they should. Drafting, triage, chasing, reconciliation — the work nobody has time for.Scoped triggers and stop conditionsA human approval step before anything leaves the buildingCost ceilings and per-request loggingRetries, dead letters and an audit trail
02Custom skillsTools the model can actually call: your APIs, your data, your rules. Not a chat box bolted onto a product.Typed tool schemas against your own endpointsPermissions per skill, per roleAn eval per skill, run in CIGraceful fallbacks when a tool is down
03AI-ready codebasesRepositories a model can navigate without guessing: typed boundaries, written conventions, and tests that catch drift before a release does.Conventions documented where agents read themTyped API contracts as the source of truthPrompt and skill versioning alongside the codeAn eval suite that fails the build, not the customer
What “AI-ready” means in the repoShipped on every AI engagement
/agentsConventions and task guides an agent reads before touching code/evalsThe inputs you expect, the five you dread, and the accepted output for each/promptsVersioned, reviewed and diffable like any other source file/skillsTool definitions, schemas and permissions, one folder per capability/contractsTyped API boundaries generated once and consumed everywhere
Evidence

How we know it will hold up.

Four things we can show you
One workspace. Shared contracts. Clear ownership.
One workspace. Shared contracts. Clear ownership.
The working relationship

Senior people. Clear decisions. No handovers.

01 · Direct access
The people in the room are the people doing the work.

Scope, design and engineering decisions stay close together—without an account-management relay.

From first conversation
02 · Visible progress
You see the product running throughout the engagement.

Regular working builds make progress visible and surface difficult decisions while they are still inexpensive to change.

Through every delivery
03 · Clean ownership
The code is yours from the first commit.

Delivery includes documentation, a support window and a codebase the next team can actually understand and run.

After handover

A small senior team, direct communication and one accountable owner from the first conversation to release.

How it works

Five steps, no surprises.

The same process whether it is a two-week spike or a six-month build.

Objectives, audience, competitors and constraints, written down before anything is designed. This is where the expensive mistakes get caught.

Wireframes then hi-fi design, covering the journey rather than a handful of hero screens. Feedback happens here, not in code.

One direction, agreed and specified. Tokens, components and the pages they compose into.

Clean, typed, tested code with a deployment pipeline from day one. You see it running every week, not at the end.

Release, handover, documentation and a support window. The team that inherits it can actually run it.
20+
Years shipping software
12
Products in the index
5
Global brands
0
Briefs lost in handover
Got an idea in mind?

Let’s start growing your business.

Tell us what you are building. If it is not something we should take on, we will say so and point you somewhere better.