What we get done — task by task
These are the areas we work in. You write tasks on your workspace board, we work through them in your order: each with acceptance criteria, delivered 48 hours after we start it (counted Mon 8 am – Fri 4 pm). The first task is free.
Workflows that actually run on their own
Every company has steps someone does by hand simply because nobody has solved them yet: moving data between tools, chasing approvals, asking for status. We build those workflows in n8n, Make or Zapier — or directly against the API when ERP, CRM and accounting need to talk to each other. Including the exceptions off-the-shelf tools choke on.
Typical tasks
- Form → CRM + Slack notification
- Shopify order → Lexoffice invoice, with error handling
- Sync HubSpot ↔ Pipedrive
- Migrating from Zapier/Make to n8n (in several tasks)
What you get
- Recurring tasks run without anyone touching them
- No more maintaining the same record in several systems
- Error handling: the workflow tells you when something breaks
- Credentials stored as credentials, not inside the workflow
Too small for an agency, too specialised for your own team
The Vercel build fails, WordPress shows a critical error, the Shopify theme misbehaves after an update, an API suddenly returns 500s. You describe the bug and invite us to your staging environment or repository; we deliver the fix as a pull request or on staging — with an explanation of the root cause. If the cause can't be pinned down without analysis, we start with a diagnosis and then put the fix in your backlog as a task.
Typical tasks
- Fix a Next.js “Hydration failed” error
- WordPress: critical error after a plugin update
- Vercel build failing after a framework upgrade
- Adjust a Shopify theme without extra apps
What you get
- Fix delivered as a pull request — traceable and reversible
- Work happens on staging or a branch, not live in production
- Root cause explained, not just the symptom patched
- Unclear cause? Diagnosis first, then you decide on the fix
Small development tasks, handed over properly
Not every extension needs a project. An extra field with validation, a CSV export, a script that reconciles data every morning, a new endpoint for the partner API: we build tasks like these in your existing code — on a separate branch, with tests, as a pull request you review and merge. Anything bigger we split with you into several tasks, each with its own acceptance criteria.
Typical tasks
- Script for a recurring task (Python or Node.js)
- New feature in an existing web app
- REST API endpoint with validation and documentation
What you get
- Built in your repository, not in someone else's system
- Separate branch and pull request — you decide whether to merge
- Tests for what's new
- Short documentation in the code and the pull request
No more retyping
Invoices, delivery notes, contracts, forms, email attachments: information arrives unstructured and gets typed into a system by hand. We extract those documents with OCR or AI, check the values against your rules, and hand them over as structured data — as a spreadsheet, into your accounting, or into your line-of-business system.
Typical tasks
- Fill a PDF form automatically from a spreadsheet
- Extract email attachments with AI → Google Sheet or DATEV folder
- Replace an Excel macro with an automation
- Extract and validate supplier invoices, then pass them to accounting (in several tasks)
What you get
- Capture without manual data entry
- Automatic validation against master data and rules
- Uncertain cases go to review instead of failing silently
- Works with scans and inconsistent layouts too
Assistants that know your processes, not just how to chat
A generic chatbot answers questions nobody asked. We wire language models in where they take work off your hands: an assistant that answers from your documents, an AI feature inside your existing software, classification of incoming requests. And we'll tell you honestly when plain automation is cheaper.
Typical tasks
- Categorise incoming requests with AI
- Integrate the ChatGPT or Claude API into existing software
- AI assistant on your own documents (RAG, in several tasks)
What you get
- Answers grounded in your data, not the model's general knowledge
- Clear boundaries: it hands over instead of guessing
- AI disclosure under the EU AI Act (Art. 50) built in
- Model choice made with data protection and running costs in mind
When it's more than a task
A task is one contained result that can be finished, checked and handed over within 48 hours (counted Mon 8 am – Fri 4 pm) — for example a bug fixed, an interface connected, a workflow built, a report automated. Larger projects we split with you into several tasks, each with its own acceptance criteria. As soon as a place is free, we start the top card in your backlog — on the next working day at the latest.
Task by task in the Prozess-Abo
You collect tasks in the backlog and set the order. In Prozess-Abo 2,5 one task is in progress at a time, up to four a month; in Prozess-Abo 5 one at a time with no monthly limit, in Prozess-Abo 10 two. Cancel monthly, no minimum term.
See the Prozess-Abo →First, find out what fits
Questions before you start, or more than two tasks in parallel? In the free Projekt-Check (30 minutes) we go through your project — and, if needed, make you an offer for an individual Prozess-Abo.
Book a Projekt-Check →Start with one task — the first one is free.
The first task is free — one per company, businesses only. It has to be doable within 48 hours (counted Mon 8 am – Fri 4 pm); we check that before we start. Afterwards you decide whether to order a Prozess-Abo. It never starts automatically.