AI and machine learning for SMBs and startups · Berlin
Machine learning research, deployed into production.
Founded at Freie Universität Berlin in 2024, we help companies with research and development in AI and machine learning. We advise on where AI and machine learning pay off and build systems that run in production.
Published case study
A RAG chatbot that cuts pre-CT consent time by 38%.
PIA answers patients' questions before a CT scan, grounded in validated clinical documents. We tested it in a prospective comparative study with 86 patients.
PIA in a hospital-information deployment. The study version covered pre-CT consent.
120 s vs 195 s, p = 0.04
vs 8.93 control, not significant
vs 8.86 control, not significant
43 PIA / 43 control, IRB-approved
Much of the consent information is the same for every patient. PIA delivers that part, and patients understand it as well as in the physician consultation.
Follow-on projects in obstetrics and surgical informed consent are under way in Leipzig, Berlin and Munich.
Busch, Kaibel, Nguyen et al., Journal of Imaging Informatics in Medicine 38, 4312–4323 (2025) · doi.org/10.1007/s10278-025-01483-w
What we build
Two service lines, one research team.
Clarity before commitment.
We help you find where AI creates real value, build the roadmap, and stay involved to make sure it delivers.
- Opportunity discovery and evaluation
- Roadmapping and governance guidance
- Interim AI leadership
- Workshops for engineering and leadership teams
- HR and recruiting support for ML roles
From prototype to production.
We build the systems: MVPs, full-stack pipelines, data analysis, dashboards. Computer vision, NLP, time-series.
- LLM document search and chatbots
- Object detection and counting in images
- Time-series forecasting and anomaly detection
- Predictive analytics and dashboards
- Evaluation, explainability, deployment, monitoring
The CIATA process
Three phases, from understanding your business to a system in production. Evaluation and documentation are always included.
We also train your team.
Workshops for engineering and leadership teams, from ML fundamentals to evaluating AI investments.
WorkshopsWe also train your team.
Workshops for engineering and leadership teams, from ML fundamentals to evaluating AI investments.
WorkshopsFrom our projects
Selected work.
Predictive analytics
Project planning: predicting success probability with statistical models.
Predictive maintenance
Data analysis: predictive maintenance and error detection.
Computer vision
Image and video analysis: flow analysis and detection of specific blood cell variants.
The team
Five researchers, dozens of projects.
Our research spans computer vision, NLP and robotics.
Prof. Dr. Tim Landgraf
Managing Partner
David Dormagen
Lead ML & Partner
Lukas Kaibel
ML Engineer
Manuel Heurich
ML Engineer
Max Granz
ML Engineer
Prof. Dr. Tim Landgraf
Managing Partner
David Dormagen
Lead ML & Partner
Lukas Kaibel
ML Engineer
Manuel Heurich
ML Engineer
Max Granz
ML Engineer
Engagement models
Defined deliverables, fixed timelines.
4–6 weeks
Strategy & governance sprint
Opportunity analysis, structured proposal, signature-ready roadmap with governance guidance.
3–6 months
Delivery & pilot phase
Production-ready pilot or live solution with measurable KPIs.
Ongoing
Executive sparring
Advisory, interim leadership, and recruiting bridge for AI talent.
8–12 weeks
MVP acceleration
Validated MVP built by applied ML researchers, production-ready.
Equity model
Tech-for-equity
Cash fee plus a small equity stake. Aligns incentives for early-stage teams that need to move fast.
Ongoing
CTO sparring
Technical leadership and recruiting bridge for ML talent.
