Skip to content
Nibble Square

Services

Five ways we get involved

Most engagements combine a few of these. If you are not sure which one you need, that is usually a sign the first conversation should be about the problem rather than the solution.

We build for clients the way we build for ourselves

Passlay is a live product with real organisers running real events on it. We designed it, built it, and we operate it — including the unglamorous parts: on-call, migrations, refunds, and the traffic spike when tickets go on sale. That experience is what we bring to client work, and it is why our estimates tend to include the parts other teams forget. See what Passlay does.

Custom software development

Systems built around how your business actually operates, not around a template.

Some problems do not have an off-the-shelf answer. For Sahti, a French healthcare practice, that meant a practice management system covering scheduling, patient records and doctor workflows in one place — replacing a patchwork of spreadsheets and paper. We start by learning the workflow properly, then build the smallest thing that replaces it, so the software matches how the team already works rather than the other way round.

TypeScriptPythonPostgreSQLDjangoNext.js

What that usually includes

  • Discovery and technical scoping
  • Domain modelling and data architecture
  • Application build and rollout
  • Migration from spreadsheets or legacy systems
  • Handover documentation and team training

Product engineering

An embedded team that takes a product from first sketch to something with real users on it.

This is how Passlay got built, and it is the work we know best. We join as a product team rather than a set of hands: shaping scope, making the architectural calls, shipping in increments you can put in front of users, and staying on past launch to fix what the first cohort actually complains about.

Next.jsReactNode.jsPostgreSQLStripe

What that usually includes

  • Product discovery and scope shaping
  • Architecture and technical direction
  • Iterative delivery in two-week cycles
  • Instrumentation and usage analytics
  • Post-launch iteration and support

Web & mobile applications

Fast, accessible web apps and installable, mobile-first builds that share one codebase where it makes sense.

Interfaces are where most software is judged, so we treat the front end as engineering rather than decoration. For Zee Soft we built a mobile-first platform that installs and behaves like a native app while shipping from a single PWA codebase — no separate iOS and Android builds to keep in sync. That means measurable performance budgets and accessibility checked against WCAG rather than assumed, on web or mobile.

ReactNext.jsTailwind CSSPWAVercel

What that usually includes

  • Design system and component library
  • Responsive web application build
  • Installable, mobile-first (PWA) builds
  • Accessibility audit and remediation
  • Performance budgets and Core Web Vitals work

Cloud, DevOps & integrations

The infrastructure and plumbing that keeps everything above running without drama.

Deploys should be boring. For Dev Geon we rebuilt their AWS architecture as infrastructure as code, turning changes that used to need a support ticket into a config change. That is the same approach we bring to CI/CD, observability, and the third-party integrations — payments, identity, messaging, ERP — that production software inevitably needs. Existing setups are welcome; we are as happy stabilising something as building it fresh.

AWSTerraformDockerGitHub ActionsGrafana

What that usually includes

  • Infrastructure as code and environment setup
  • CI/CD pipelines and release automation
  • Monitoring, logging and alerting
  • Third-party and payment integrations
  • Cost review and scaling plans

AI & machine learning

Applied models wired into real products — with evaluation, guardrails and a fallback when they get it wrong.

We build AI features that survive contact with production. For Safe and Fit's CEO, that was Trial Rocket — an OpenAI-based chat agent that drafts reply suggestions and sends messages, including WhatsApp, on her behalf. The engineering around the model matters more than the model itself: evaluation sets, cost controls, latency budgets, and a sensible answer for what happens when confidence is low.

PythonOpenAILangChainpgvectorNode.js

What that usually includes

  • Feasibility review and data assessment
  • Retrieval and RAG pipelines
  • Model integration and prompt architecture
  • Messaging and WhatsApp integrations
  • Evaluation harnesses and quality benchmarks

How an engagement runs

What working together looks like

Same four steps whichever service you start from.

  1. 01

    Understand

    We sit with the people doing the work, map the current process end to end, and find where the real cost is. Usually it is not where anyone expected.

  2. 02

    Shape

    We turn that into a scoped plan with an architecture, a sequence and honest estimates — including what we would cut first if time gets tight.

  3. 03

    Build

    Two-week cycles, working software at the end of each one. You see progress continuously rather than at a milestone review.

  4. 04

    Run

    We launch, watch what real usage does to our assumptions, fix what breaks, and hand over something your team can own.

Not sure which of these you need?

Describe the problem and we will tell you what we think it actually is — including if the answer is that you do not need us.

Start a conversationRead testimonials

Currently booking projects starting next quarter.