Skip to content
GroundspeedStudio

We build software that moves your business measurably forward.

An independent studio for product development, software architecture and applied AI. We work AI-first: language models go where they create real leverage — and stay out where they would only add risk.

Start a projectSee our work
years building products
12+years building products
projects shipped
40+projects shipped
average time to launch
6 wksaverage time to launch

Selected clients

  • Voltaris Energie
  • MEDIRA Health
  • Nordlicht Logistik
  • Fintura
  • Hafenwerk Immobilien
  • SOLARA Group

Services

Four things we are genuinely good at

We keep our offering deliberately narrow. Everything listed here we have been doing for over a decade — anything else we happily refer elsewhere.

01

Product, web and app development

From the first sketch to running in production. We design and build web applications, iOS apps, internal tools and customer portals that are still maintainable three years later — with a team that understands the product, not just the ticket.

  • Web apps and portals with Next.js, React and TypeScript
  • Native iOS apps in Swift and SwiftUI, through to App Store release
  • Interface design — from the first draft to a documented design system
  • Integration with existing ERP, CRM and auth systems
  • Accessibility and performance treated as requirements, not extras
02

Applied AI & AI engineering

We build AI features that survive daily use: clear boundaries, traceable answers, and a way to measure whether the system is getting better or worse. No demo magic — production systems.

  • Assistants and search over your own data (RAG, vector search)
  • Agentic workflows for repetitive domain processes
  • Evaluation and monitoring: measurable quality instead of gut feel
  • Privacy-first architecture, including regulated industries
03

Architecture & technical advice

For decisions that get expensive the moment they are wrong. We review your system, name the risks in plain language and propose a path that fits your team — not our portfolio.

  • Code and architecture audits with a prioritised action plan
  • Legacy migrations in stages rather than one big bang
  • Second opinion before a tender or a rewrite
  • Technical sparring for founders and product teams
04

Operations & continuous delivery

Launch is the beginning, not the finish line. We stay on as a reliable point of contact — with fixed monthly capacity, agreed response times and a backlog we prioritise together.

  • Fixed monthly capacity instead of vague timesheets
  • Monitoring, error tracking and performance budgets
  • Regular dependency updates and security patches
  • Quarterly review: what worked, what do we drop?

Process

How a project runs with us

No waterfall, no agile theatre. Four phases, each with a tangible result — and after every phase you can walk away with something useful in hand.

  1. 013 – 5 days

    Frame it

    Before writing a line of code we make sure we understand the problem. We talk to the people who will actually use the thing, look at the existing systems and write down what success concretely means. Often this is where it turns out the cheapest solution is a different one than planned — and we say so.

    A document with the target picture, risks, an architecture sketch and a defensible estimate.

  2. 021 – 2 weeks

    Prototype

    We build the riskiest part first, not the easiest. You get something clickable that tests the project's central assumption. On AI projects the evaluation comes first: we measure how well the system handles real cases before building a product around it.

    A working prototype on real data, plus an honest read on what holds up and what does not.

  3. 032-week cycles

    Build

    We ship to a staging environment every two weeks, open to you at any time. Each cycle ends with working software, not a status report. You follow progress on a shared board and always know where the budget stands.

    A usable build every two weeks, test coverage on the critical paths and a current budget status.

  4. 04ongoing

    Operate

    After launch we hand over properly: documentation, a handover workshop, access to everything. If you want, we stay on with fixed capacity and keep building. If your team takes over, we make sure it works without us.

    Documentation, a handover workshop, monitoring — and an operating model that fits your team.

Work

Selected projects

A cross-section of recent years — from a customer portal for an energy provider to an AI assistant running inside a hospital.

See all work

Principles

What we hold to

A handful of convictions formed over years. They explain reasonably well why working with us is occasionally uncomfortable and usually productive.

AI is a tool, not the point

We work AI-first, which means we genuinely check whether a model solves a problem better than code. Often the answer is yes, and then we build it. Sometimes the answer is no, and we say so — even when the AI project would have been the larger contract.

We build for the day after launch

Software is maintained for far longer than it is written. So we favour boring, well-documented technology, write tests for the paths that hurt, and leave behind code someone else can pick up.

One point of contact, no handoffs

You talk to the people who actually build the product. No relay from sales to a project manager to a team you have never met. That caps how big we can get — which is precisely the point.

Transparency, especially when it stings

Budget status, risks, misjudgements: you hear about them as they happen, not in the closing report. We have yet to see a project where bad news improved by being held back.

Technology

What we work with

We pick boring, well-documented technology wherever we can, and new tools only where they buy something real. This is not a certificate collection — it is what actually runs here day to day.

Frontend & apps
TypeScriptReactNext.jsAstroSwiftSwiftUITailwind CSS
Design
FigmaDesign tokensStorybookPrototypingWCAG 2.2
Backend & data
Node.jsPythonPostgreSQLDrizzle ORMRedistRPC
AI & machine learning
Claude APIOpenAI APIMCPpgvectorLangGraphEvals
Infrastructure
VercelAWSDockerCloudflareTerraformGitHub Actions
Quality & operations
VitestPlaywrightSentryOpenTelemetryLighthouse CIXCTest

Working together

Three ways to start

Which model fits usually becomes clear in the first conversation. Prices are planning figures — a binding quote follows the framing call.

Framing sprint

from €4,500

1 week

For work where it is not yet settled what should be built. You end up with a defensible target picture, an architecture sketch and an estimate you can work with internally — even if you build it with someone else afterwards.

  • Interviews & system review
  • Architecture sketch
  • Effort estimate
  • Risk register
Start a project

Product development

★

from €18,000

from 6 weeks

The usual case. We build in two-week cycles and you see a usable build after each one. Billed either as a fixed price per increment or time-and-materials with an agreed ceiling.

  • Two-week cycles, each ending in something usable
  • Staging environment from day one
  • Tests on the critical paths
  • Handover workshop included
Start a project

Retained capacity

from €3,200 / month

cancel monthly

For products under continuous development. You book a fixed number of days per month and we prioritise together. No ticket ping-pong, no twelve-month lock-in.

  • Fixed days per month, prioritised together
  • Committed response times
  • Maintenance, updates and security patches
  • Quarterly review
Start a project

Indicative, excl. VAT. Binding quote after the first call.

Common questions

Before you ask

How big is the team, really?

Small, and deliberately kept that way. We do the core of a project ourselves; for specialist work such as design, accessibility audits or load testing we work with a settled circle of partners we have known for years. You will know who is working on your project before it starts — not when the invoice arrives.

Do you take over existing projects?

Yes, and it is a good share of what we do. In those cases we start with a short audit so both sides know what they are getting into. If we conclude that taking it over does not make sense, we say so after the audit — and you still walk away with a solid assessment of your system.

What does “AI-first” actually mean here?

That we seriously check, for every problem, whether a language model solves it better than written code — and that we use AI tooling internally, which noticeably shortens our delivery times. It explicitly does not mean every product needs a chat window. Where a model would only add uncertainty, we leave it out.

Who owns the code?

You do, entirely, from the first line. You get access to the repository from the start, not at the end of the project. We work inside your organisation or hand everything over in full — we do not build in dependencies on ourselves.

Do you work remotely or on site?

Mostly remote, with fixed on-site dates at kickoff and handover if that matters to you. For clients across German-speaking Europe this has been the norm for years and works well — as long as the dates are firm.

Contact

Tell us what you are planning

A short description is enough. We reply within one working day and tell you honestly whether we are the right fit — and if not, we usually know someone who is.

hallo@groundspeed.studio

Typically a reply within 24 hours