Software your business runs on.


We design and build web platforms, mobile apps and the infrastructure underneath them. Then we stay and keep them running.

ScrollTaking on new projects
Reply within one working dayFixed price after discoveryCode and infrastructure in your accountsA staging URL from week oneRunbook with every systemThirty days of support after launchNo lock-in

— The studio

We are the team you call when the product has to ship, and keep shipping. Small enough to answer the phone. Senior enough to own the whole stack, from the first wireframe to the on-call rota.

  • ReplyOne working day
  • OwnershipIn your accounts
  • PricingFixed, after discovery
  • Core stackTS · Go · Postgres

01 — Capabilities

What we build, and keep running.

Six things we do well enough to put our name on. Everything else we will tell you to buy, or point you to someone better.

  1. 01

    Custom software

    Internal tools and back-office systems.

    Built for the people who use them every day. We start from their workflow, not a framework, and do the boring parts properly: permissions, audit logs, exports.

    • TypeScript
    • Postgres
    • Go
  2. 02

    Web platforms

    Marketplaces, portals and SaaS products.

    Astro or Next.js on the front, Postgres underneath, nothing you do not need. Fast by default and measured in the field, not only in Lighthouse.

    • Astro
    • Next.js
    • Node
  3. 03

    Mobile apps

    One codebase where it fits, native where not.

    Shipped through both stores, instrumented from the first build, maintained after launch. We handle the store review so you do not have to learn it.

    • React Native
    • Swift
    • Kotlin
  4. 04

    Cloud & DevOps

    Infrastructure as code, CI/CD, observability.

    Boring, reliable, documented. Mostly AWS and Hetzner, always Terraform, always a runbook. You get the keys, the diagrams and the alerts.

    • Terraform
    • AWS
    • Hetzner
    • Grafana
  5. 05

    AI & automation

    Language-model features that earn their place.

    Document processing, support triage, search. Evaluated against your real data before they ship, with a fallback for when the model is wrong.

    • Claude
    • Evals
    • Pipelines
  6. 06

    Product design

    Research, flows, interface, design system.

    One designer embedded in the engineering team, so the design that ships is the design that was made. Clickable prototypes before production code.

    • Research
    • Prototyping
    • Design systems

02 — What we commit to

  1. 1 dayTo a reply from a person who would work on your project
  2. 6 wkTarget from kickoff to a first version in production
  3. 100%Of code, infrastructure and domains in your accounts
  4. 30 daysOf support after launch, included in every project

03 — How we work

Six rules we do not negotiate.

They are the reason projects with us are calm. They are also the reason we occasionally say no.

  1. 01

    The runbook before the feature

    Every system ships with the document that explains how to run it at three in the morning. If we cannot write it, the feature is not done.

  2. 02

    A staging URL from week one

    You see real software early and often. Not slides, not mockups of software.

  3. 03

    You own the repository

    Code, infrastructure, domains, keys: in your accounts from the first commit. Leaving us should be easy, which is why nobody does.

  4. 04

    No hidden work

    Every hour is on a board you can open. Estimates come with the assumptions attached.

  5. 05

    Boring infrastructure

    Postgres, Terraform, well-known cloud primitives. Exciting technology goes into products, not into the platform they stand on.

  6. 06

    Measured in the field

    Performance budgets, error rates and user timings are tracked in production. Lighthouse scores are a starting point, not a result.

04 — Process

How a project runs.

  1. 01 · 1–2 weeks

    Discover

    We talk to the people who will use it. We read the old code. We write down what done means, and what it costs.

  2. 02 · 2–3 weeks

    Design

    Flows first, screens second. A clickable prototype before a line of production code, tested with real users.

  3. 03 · Two-week cycles

    Build

    A demo every Friday. A staging URL you can send to anyone. Tests, monitoring and documentation as part of the work, not after it.

  4. 04 · Ongoing

    Run

    Monitoring, on-call, upgrades. We stay for as long as the system runs, and hand over cleanly if you ever want to leave.

05 — Architecture

What a typical system looks like.

The shape of a typical system we build and run. Every block is a part we take responsibility for, from the first commit to the on-call rota.

Reference architecture · Sheet 01Hover a block to trace its connections
01 · Clients02 · Edge03 · Services04 · DataWeb appAstro · Next.jsA1Mobile appReact NativeA2Back-officeInternal toolsA3Partner systemsERP · Payments · CRMA4CDN · WAFCloudflareB1API gatewayAuth · Rate limits · VersioningB2Webhooks inSigned · ReplayableB3Core APIGo · NodeC1Background jobsQueues · SchedulesC2Event busAt-least-onceC3LLM workersClaude · Evals · FallbacksC4PostgresSystem of recordD1RedisCache · LocksD2ClickHouseAnalyticsD3Object storageFiles · ExportsD405 · PlatformInfrastructure & observabilityTERRAFORM · AWS / HETZNER · GITHUB ACTIONS · GRAFANA · SENTRY · ON-CALL

06 — Stack

Chosen for how they behave in year three, not for how they demo in week one.

  • Frontend
    • TypeScript
    • Astro
    • Next.js
    • React
    • Vue
  • Mobile
    • React Native
    • Expo
    • Swift
    • Kotlin
  • Backend
    • Node
    • Go
    • Python
    • GraphQL
    • REST
  • Data
    • Postgres
    • ClickHouse
    • Redis
    • S3
  • Infrastructure
    • Terraform
    • AWS
    • Hetzner
    • Cloudflare
    • Docker
    • GitHub Actions
  • Quality & AI
    • Playwright
    • Grafana
    • Sentry
    • Claude
    • Evals

07 — What you get

Every project ships with the same eight things.

Not extras, not a premium tier. This is what done means here, whether the project takes six weeks or a year.

  1. 01

    Code in your repositories

    GitHub or GitLab under your organisation, from the first commit.

  2. 02

    Infrastructure as code

    Terraform for everything that runs. Rebuildable from scratch.

  3. 03

    Documentation and runbooks

    Architecture, decisions, and what to do at three in the morning.

  4. 04

    Monitoring, alerting, on-call

    Dashboards, error tracking and a rota. You see what we see.

  5. 05

    CI/CD with staging

    Every branch gets a preview. Every release takes the same road.

  6. 06

    Automated tests

    Unit, integration and end-to-end where it pays. Run on every release.

  7. 07

    A recorded handover

    A walkthrough of the system for your team, kept for the next hire.

  8. 08

    Post-launch support

    Thirty days after launch are on us. Then a retainer, if you want one.

  • Fixed price after discoveryThe number comes with the assumptions written next to it.
  • No lock-in, everLeaving us takes a handover call, not a migration project.
  • Reply within one working dayTo every message, from a person who works on your project.

08 — When to call us

The first version worked. Now it has real users, and it has to be rebuilt without stopping for a day.
Situation 01 — Rebuild without downtime

09 — Questions

Things people ask before they write.

  1. You do, from the first commit. Repositories, infrastructure and domains live in your accounts; we only hold access while we work.

  2. Fixed-scope work is priced after a paid discovery phase, with the assumptions written down next to the number. Ongoing work is priced monthly.

  3. Yes, before the first call if you prefer. We have a standard mutual NDA and are happy to review yours.

  4. Usually. We start with a two-week audit: code, infrastructure, costs and risks. You get a written report whether or not you continue with us.

  5. In the EU, working in CET. Overlap with North American mornings is normal for us.

  6. You keep everything, we hand over documentation and access, and the retainer ends at the end of the month. No exit fees.

10 — Contact

Tell us what needs to work.

01 — By email

hello@codeglow.eu
Reply within one working day

Send a few lines about the project, the deadline and the budget range. We reply within one working day, with questions rather than a quote.

02 — Or write here

Both land in the same inbox

  1. 01We replyWithin one working day, with questions rather than a quote.
  2. 02A 30-minute callScope, constraints, timeline. You talk to the engineer who would lead the work.
  3. 03A written proposalWithin a week. Assumptions, price, plan. Valid for 30 days.