Success Story
LLM Contract Review
Vitesco develops an LLM-based contract use case, setting new AI standards across the company.
#Automobile / Mobility
#Business Analysis & Concept Development
#Azure
The Customer Need
More efficiency in contract handling. Quick insights instead of drawn-out analysis. As part of a broader process optimization effort, we defined the use case “Automated Contract Analysis.” Vitesco Technologies wanted to find out if AI could simplify the demanding legal contract reviews. At that time, the company had no hands-on experience with AI—plus, the data involved was highly sensitive. Choosing the right LLM was critical. After thorough assessments and tests, the open-source model Mistral was selected in October 2023.
Challenges & Extra Miles
Missing standards & benchmarks. Contracts come in many forms—with varied structures, layouts, and contexts. Our first challenge was to use an LLM to develop a structure that made these contracts machine-readable and analyzable.
Innovation
The smart contracting recipe. | Start with a vector database (Qdrant), add a large language model, a dedicated user interface, and blend it all with compressed expert knowledge – deployed via a trusted hyperscaler. Then comes the real magic: custom AI training to inject industry-specific knowledge tailored to Vitesco Technologies.
The result? A smart tool that:
Stores domain-specific company knowledge
Is accessible to all expert users
Filters and displays contracts effectively
Features semantic search to detect special clauses in legacy contracts
Enables chatbot-style interactions via Natural Language Processing (NLP)
Offers personalized contract overviews
Flags risky clauses to mitigate legal exposure
Includes an admin panel for custom settings
Supports structured search across supplier/customer contracts
Big Wins
Laying the claim. | Just a short while ago, generative AI was still a frontier for most of us. Vitesco Technologies approached it with curiosity and courage. Today, their contract management team has become the company’s go-to unit for all things GenAI.
Up to 80% reduction in contract review time
Less than 3 months to MVP completion
40.000 + contracts now integrated into the solution
Small Victories
Prompt Engineering Training. LLMs live on prompts. But what makes a prompt “work”? Our expert delivered training to establish foundational understanding, equipping Vitesco’s teams to better evaluate and craft future AI use cases.
Better done than perfect. The “fail forward” mindset often beats perfectionism. It takes a cultural shift - but once that’s in place, like here, results start coming fast.
Next Steps & Future Upgrades
Always room for better. Next up? Features like automatic clause rewording, contract comparisons with previous versions, and expanded use cases as token limits evolve. This flexible setup means many more applications can be built with just minor tweaks.
"The potential of generative AI is enormous. At Vitesco Technologies, we saw real opportunities, especially in contract management. This use-case-driven approach helped us take strong first steps. The insights we’ve gained on LLMs, prompting, and automation are now driving AI adoption across the entire company."
Alina Stahl, Head of Contract Management bei Vitesco Technologies
What we delivered in this project:
IT- & Digital-Consulting
Implementation & Optimization
Operational Excellence & Support
Wollen sie auch zu einer unserer Success Stories werden?
Lassen Sie uns reden!
Kevin Heinloth
Senior Sales Specialist