Peppermint’s week at Microsoft’s AI Co-Innovation Lab Munich
Peppermint is kicking off the year with energy, innovation, and a commitment to pushing the boundaries of artificial intelligence (AI) in legal tech. With our strategic partner, Microsoft, we recently collaborated on a proof-of-concept (POC) where we were able to accelerate our AI advancements and expand our reach.
During our recent experience at Microsoft’s AI Co-Innovation Lab in Munich, we were able to delve deep into collaborative projects that highlighted our shared vision. Together, we aim to bring transformative AI solutions to legal firms, ensuring that security, responsibility, and operational efficiency are woven into the fabric of our offerings.
The Microsoft AI Co-Innovation Lab is a modern space designed to foster collaboration between industry leaders and Microsoft’s AI experts to co-develop, test, and refine AI-driven solutions that deliver real-world impact.
For the Peppermint team, this meant the possibility to explore the newest advancements in Microsoft’s AI services, enhancing LARA (Legal Analysis and Research Assistant), and addressing legal specific challenges around data privacy, compliance and responsible AI adoption.
Laying the Foundations with Large Document Understanding
Peppermint’s Chief Innovation and Technology Officer, Mike Walker, together with Rich Mephan and Ben Williams joined Microsoft AI experts, spending a week working on improving and accelerating our platform’s AI capabilities and understanding.
One of the biggest challenges in legal AI is understanding how to extract meaningful insights from large, complex legal documents. As Mike Walker explains, “This is something fundamental that all legal software companies need to understand to produce a measurable solution for legal businesses.
From the beginning, we looked at the architectural challenges of working with large documents in LLMs, extracting intent, insights, and key information efficiently, and more importantly how can AI assist in contract analysis and decision-making support. This way we were able to establish a strong baseline for our next steps.”
Engineering the AI-Driven Legal Workflow
“By having a strong foundation, we moved from concept to implementation”, says Mike. This meant trialing different orchestration tools to evaluate differences from LangChain to Semantic Kernel, which allowed Peppermint to optimise AI orchestration across different legal workflows, improve performance and interoperability with Microsoft AI ecosystems and enhance contextual awareness within complex document processing.
The team then built on top of this orchestration layer. “By grounding our LLMs with contextual business data, we obtained new ways to improve accuracy in legal decision-making, connect AI insights with broader case management processes and optimise efficiency in legal research and contract review,.” Mike shared.
“By leveraging these orchestration tools, we can also manage agentic flows – the process of guiding and automating tasks in a way that mimics human decision-making – to maximise the information extracted for follow-up processes, task management, risk assessment, and more,” adds Mike.
“The foundations built in our proof-of-concept (POC) allow Peppermint to seamlessly manage these flows, ensuring that every piece of extracted information is utilised effectively. This orchestration not only enhances contextual awareness within complex document processing but also enables dynamic and responsive legal workflows that adapt to the needs of legal professionals in real-time. By leveraging these orchestration tools, law firms benefit from increased efficiency and reduced operational costs. Additionally, the integration of AI capabilities ensures that legal teams can focus on higher-value tasks rather than mundane administrative duties, thereby improving overall productivity and client satisfaction."
In addition to this, the orchestration tools provide a framework to integrate AI insights with broader case management data, ensuring that data flows seamlessly across different platforms and stages of legal processes. This approach enables proactive identification of potential risks, automated generation of relevant legal tasks, and an overall improvement in the efficiency and accuracy of legal operations. Through this strategic orchestration, Peppermint is poised to deliver AI solutions that not only extract valuable insights but also translate these into tangible actions, driving significant value for our clients.
From Proof-of-Concept to Customer Value
One of Peppermint’s main objectives is to bring AI innovation into the hands of legal professionals in a way that creates real and measurable value.
We started focusing on building more complex legal process flows that drive real business value, reducing human intervention where possible while still maintaining responsible AI principles and turning document understanding into actionable workflows, where AI moves beyond insights and triggers real-world actions.
This paved the way for exciting new opportunities including AI-driven contract review, content understanding, business information review and risk assessment.
Throughout the lab, Peppermint adhered to our responsible AI principles, ensuring that every aspect of the AI capability we design complements legal professionals’ needs and allows faster and more accurate decision-making.
Key Takeaways & The Future of AI in Legal Tech
“This experience at the AI Co-Innovation Lab reaffirmed the transformative potential of AI in legal workflows. We left with a working proof of concept which accelerates our roadmap for LARA, new insights into AI orchestration at scale and a clear roadmap for customer adoption,” states Mike.
In the coming months, Peppermint will focus on bringing this AI-driven legal technology into production, gathering feedback from legal professionals to refine and improve our approach and continue our collaboration with Microsoft to push AI boundaries even further.
Most importantly, this has ignited fresh possibilities for law firms to work smarter, faster, and more efficiently. This is just the beginning—stay tuned for what’s next!