Skip to content
Felix Schumann
Get in touch

Services / Automation engineering

What repeats itselfis the machine’s job now.

Turn recurring manual work into a process you can follow. I connect your data, existing software and suitable AI models into a system your team can use every day.

See it builtDiscuss your project

AI automation · Automation engineering · Connected workflows

Explore an example

Your current workflow
Incoming01

A new enquiry

  • 01Message received
  • 02Identify the contact
  • 03Capture the request

Your task shapes the implementation.

The everyday problem

An enquiry arrives. What happens next?

Copy a message, create a contact, ask someone in the team: much of the work happens between tools. That is where I start. We take one workflow, connect its steps and make the exceptions manageable too.

The goal is a clearly scoped, verifiable workflow: collect information, organise it and move it to the right place. Consequential decisions have an approval step, and failures remain visible.

What we will work on together.

The scope follows your project.
  1. 01

    Understand the workflow

    Analysis of a concrete workflow and a scoped first implementation

  2. 02

    Connect the steps

    Integrations, data validation and traceable processing steps

  3. 03

    Run it day to day

    An interface, error handling and a handover that fit your team

From my work

See what this looks like in practice.

All projects
In daily use2026

Research and outreach automation

Find the right brand. Reach the right person. Keep the thread.

From relevant brands to the right contact and a deliberate follow-up. A daily view keeps the next steps visible.

  • Find brands from real opportunities
  • Connect research and contact sequences
  • Keep follow-ups visible
  • Next.js
  • TypeScript
  • PostgreSQL
  • +3
Explore the projectClient project
In daily use2026

Shop research and contact pipeline

From a shop directory to contacts a sales team can use.

Public shop directories become usable data: cleaned, enriched, searchable and ready to export.

  • Separate discovery and enrichment
  • Deduplicate returning shops
  • Search and export contacts
  • Python
  • Next.js
  • MongoDB
  • +3
Explore the project112,000 leads in the database
In daily use2026

Content intelligence and production platform

A custom algorithm. Clear plans. Your content agent.

A custom-built algorithm, an editorial planner and an agent tailored to your content goals. From research to production.

  • A custom algorithm for your topics
  • A planning tool for the entire workflow
  • A content agent that fits your strategy
  • Next.js
  • TypeScript
  • MongoDB
  • +1
Explore the projectTwo editorial briefings per day

Our first conversation

Show me where the manual work happens.

You do not need a finished technical specification. Your idea, an example or a look at the current workflow is enough to start.

Let us talk it through
  1. 01

    Which task repeats, and how does your team recognise a good result?

  2. 02

    Where is the data, which interfaces exist and what may the system change?

  3. 03

    Which exceptions need a person, and who maintains the workflow after launch?

You might also be wondering:

What do you need for an initial enquiry?Can you take over an existing application?Do you provide hosting and further development?

Inside the implementation

Is an AI pilot worthwhile? Evaluate effort, quality and valueTime savings, rework and output quality: define useful success criteria before turning an AI demo into a permanent business system.Long-running AI jobs: Show progress without pretending the work is doneAccepted, processing or complete: make background work, interruptions and cancellation understandable to the people using an AI workflow.AI content planning: Turn a pile of ideas into a defensible editorial choiceConnect audience needs, original expertise and production effort. Design transparent topic selection, editorial planning and approval states.Speech to text: A good AI dictation app goes beyond transcriptionFrom recording to the right text field: what workplace dictation needs and how to evaluate a speech-input workflow.Why do CRM enquiries stall? Design clear handoffsOwnership, next actions and exceptions: turn a new enquiry into a traceable sales workflow with a custom CRM.Human approvals: Where an AI agent should wait for youDesign approvals around concrete actions without overwhelming people with endless confirmation dialogs.AI cost and latency: Measure the whole workflowWhy a cheap model call can produce an expensive workflow, and how routing, concise outputs and bounded retries can improve a product.Webhooks: Turning duplicate deliveries into a single actionDuplicate events, ordering and retries: a practical design for integrations that remain dependable when networks and workers fail.From AI text to dependable data: Using structured outputs wellValid JSON does not guarantee correct facts. A practical design for extraction, business validation and understandable failure handling.AI automation starts with a clear workflowWhich workflows suit AI automation, and why data, clear ownership and reviewable results should come before choosing a model.
Felix SchumannFelix SchumannAI automation engineer

Name it, I build it.

If something here looks like the problem you have, write to me. A short description of what is not working is enough to start.

Get in touch[email protected]