Services & pricing

Defined scope. Defined price. Defined dates.

Before the first line of code, you know what the system will do, when it goes live and what it costs. I plan and build it and put it into operation at your site — and you talk to the same engineer throughout.

See pricing Describe your project — offer in 48 h

Fixed-price packages

Three package sizes, every deliverable visible

A “use case” is one well-bounded workflow — order intake from emails, say, or answer search across your contract archive. Not sure which package fits? That is exactly what the Readiness Check answers.

Small · MVP

€15,000–25,000

2–6 weeks to acceptance

One use case, live in production — not a throwaway prototype
Agent or knowledge search built on your real data
A web interface your team starts using immediately
Runs via cloud API or on your servers
Documentation and two weeks of post-launch support

Ask about Small

Most chosen Medium · Production

€25,000–50,000

6–12 weeks to acceptance

One complete business process, automated end to end
Agents operate your systems and coordinate the steps
Model adaptation to your domain language, or advanced knowledge search
Automated tests and quality monitoring run alongside
Four weeks of post-launch support and tuning

Ask about Medium

Enterprise · Platform

from €50,000

12–24 weeks to acceptance

Several agents working together across processes
Custom models where standard ones fall short
Deployment up to fully isolated environments
Single sign-on, role-based access, compliance review
Three months of dedicated post-launch support

Ask about Enterprise

What moves the price within the range: the number of systems to connect, the state of your data, the approval steps required, and whether a standard model suffices or needs adaptation. The offer itemises this. In every package: source code and usage rights transfer fully to you.

Process

From check to acceptance — with a payment plan

No advance into the unknown: you pay at milestones, and acceptance follows criteria written into the offer.

Commission the build within 3 months of the check and the full check fee is credited against the project price. The acceptance criteria — measurable conditions for “done” — are written into the offer, not the fine print.

“You work with the engineer who writes the code. No account manager, no handover, no junior learning on your project.” Lukas Friedrich · founder of Tippel

Services

What I build

Four building blocks — on their own or combined in one project.

Agents that execute workflows

Software assistants read documents, check them against your data, draft replies and trigger the next step in your existing systems. Sensitive decisions go to a person for approval — every action lands in a log your compliance team can read.

For the IT team

Orchestration with LangGraph/LangChain or AutoGen; models via API (Anthropic, OpenAI, Azure OpenAI) or self-hosted behind an OpenAI-compatible vLLM endpoint — switching is configuration, not a rebuild. Integration with existing systems via REST APIs or direct database access; structured outputs with schema validation, so downstream systems receive reliable data. Guardrails and human-in-the-loop approvals at the points you define; every action lands in an audit-proof log. Delivered as Docker containers, with Kubernetes manifests on request.

Answers from company knowledge

Your employees ask in plain language and get answers with the exact source — respecting access rights. Answer quality is measured, not asserted: before acceptance you see the hit rate on your real documents.

For the IT team

Retrieval-augmented generation (RAG) with a vector database and hybrid search — semantic plus keyword; chunking and metadata per document type instead of one-size-fits-all. Permission mapping onto your existing roles and groups: anyone not allowed to open a document will not get it as an answer either. Every answer carries its source at document and section level. An evaluation pipeline with reference questions from your domain — retrieval quality is measured before acceptance and re-checked after every update; an ingest pipeline keeps the index current as documents change.

Models that speak your language

Where standard models don’t know your terminology, forms or norms, I adapt models to your documents — and demonstrate the improvement against the agreed criteria before delivery.

For the IT team

LoRA/QLoRA fine-tuning of open models with PyTorch; the base model is chosen by benchmark on your data, not by marketing. Training and evaluation data are prepared from your own stock with documented steps; models, datasets and prompts are versioned, every training run reproducible. Delivered quantised for your GPU servers or as a vLLM deployment — with a before/after measurement against the agreed criteria.

Image and sensor data

For manufacturing and engineering, I train models on your image, 3D and measurement data — for in-line quality inspection, for example. Results reach production in seconds.

For the IT team

PyTorch training on GPU and HPC infrastructure (CUDA), from classification and segmentation to 3D volume data. Inference optimised for where it runs — quantised at the edge next to the line, or on a GPU server in the data centre. Integration with control systems and existing software via common interfaces such as REST or OPC UA; latency and detection rates are measured and recorded in the acceptance report.

Where your AI runs

Three paths — chosen by requirements, not ideology

I set up all three paths GDPR-clean, with data processing agreements and documented data flows. The difference lies in time to start, cost profile and control:

	Cloud API	EU cloud, self-hosted	Your own servers
Models	OpenAI, Anthropic or Microsoft Azure	Open models in a German/EU data centre	Open models on your hardware — up to fully air-gapped environments
Time to start	Days	Weeks	Weeks, plus hardware procurement
Cost profile	Usage-based, no own hardware	Monthly server costs, predictable	One-off hardware, low running costs
Your documents	Stay in your systems; only the individual request goes to the provider, encrypted (DPA, EU endpoints, no training on your data)	Stay entirely in the EU data centre	Never leave the building
Typical choice for	Fast start, variable load, standard confidentiality requirements	US providers ruled out, predictable costs	Trade secrets, classified material, regulated environments

The paths can be mixed: sensitive documents stay in-house while uncritical load runs through the cloud. Which variant fits your case is settled by the Readiness Check — including an operating-cost estimate, not just a build price. Contracts and data flows in detail: Procurement & IT.

After go-live

Optional maintenance: so the system still delivers in month twelve

Commissioning is part of every package. Ongoing operations stay with you — or we put together a maintenance package:

Monitoring & quality

In the maintenance package, answer quality is measured continuously. If the system drifts, I see it before your users do — and fix it.

Updates & evolution

Model and security updates, regression tests after every change, and a monthly engineering allowance for new requirements.

Maintenance package from €1,500/month

Response times and availability are set in the retainer contract — plus a quarterly review with the system’s quality metrics. No retainer, no problem: you receive the system fully documented and take over yourself.

Describe your project — get a real price

A short project description is enough. Within 48 hours I reply with a fixed-price offer, or with the two or three questions still needed for one.

Request an offer Idea still fuzzy? Start with the check