Transforming Legal Tech with AI

Cicero AI’s Innovation Journey in Collaboration with the AICC

About the Business

Cicero AI is modernising legal services with AI tools designed specifically for legal professionals. Its flagship product, an AI-powered chatbot, supports legal research and automates routine tasks. By integrating seamlessly into solicitor workflows, Cicero AI enhances accuracy, boosts efficiency, and helps deliver faster, smarter legal solutions.

But legal language is nuanced, and context is everything. To deliver more relevant, jurisdiction-aware responses, Cicero AI partnered with the AI Collaboration Centre to enhance their tool’s precision, tone, and contextual domain specific awareness.

The Challenge

Cicero AI aimed to move beyond generic chatbot responses by improving the legal tone of its outputs, in addition to delivering more context-specific answers for complex legal queries, and exploring a scalable, low-code solution that could be easily used by non-technical legal professionals.

But legal AI comes with unique challenges:

  • Minimising AI “hallucinations” in sensitive legal contexts
  • Ensuring data security and trust
  • Maintaining jurisdictional accuracy

The AICC’s Transformer Programme offered an opportunity to develop a proof of concept that combined fine-tuning of a large language model with Retrieval-Augmented Generation (RAG), boosting both tone and relevance. The programme also enabled knowledge transfer and upskilling for the Cicero AI staff, ensuring they were comfortable with the approaches used, enabling them to take the project forward post transformer programme engagement.

Image for The Challenge

The Collaboration

Over 18 days of hands-on, fully funded support, the AICC team worked closely with the Cicero AI team, remotely and speedily, to shape a technical roadmap grounded in their real-world needs.

  • Fine-Tuning a Legal Model: The team formatted conveyancing data into JSONL, a standard for training AI models, and used it to fine-tune an LLM model, giving it a more “legalistic” tone and improving answer consistency.
  • RAG for Better Context: The AICC developed a Retrieval-Augmented Generation (RAG) proof of concept that pulls relevant legal content into each prompt, significantly boosting the relevance and factual grounding of responses.
  • Low-Code Deployment: Recognising Cicero AI’s preference for a no-code/low-code environment, AICC selected LangFlow, built on the LangChain framework and hosted via AstraDB.
Image for The Collaboration

The AI Solution

The final solution was a customisable, future-ready AI platform that combines:

Legal domain fine-tuning

Training a LLM with domain-specific Q&A allowed the chatbot to respond with greater consistency, authority, and a tone that reflects real legal drafting. This not only improved quality but gave Cicero AI full control over how their model communicated.

Contextual accuracy through RAG

By embedding relevant legal documents into each user query, the system produced more informed, jurisdiction-aware answers, reducing hallucinations and increasing user trust.

Low-code implementation with LangFlow

The chosen platform offered seamless integration with a current provider while allowing, for future vendor flexibility, high data compliance, and a scalable path forward. Better still, it was intuitive enough for non-technical teams to manage independently.

Full ownership

Before wrapping up, the AICC provided detailed documentation, security handover guidance, and training, giving Cicero AI full ownership of the proof of concept, and the knowledge to build it further.

Impact and Results

The solution delivered by the AICC has equipped Cicero AI to develop a robust, future-ready platform tailored for the legal sector. Seamless integration, multi-vendor support to avoid lock-in, and strong data compliance standards make it ideal for operating within a highly regulated industry. The platform is also scalable, cost-effective to prototype, and can be deployed locally or in the cloud, offering maximum flexibility as the business grows.

With a fine-tuned legal tone and smarter, context-aware responses powered by Retrieval-Augmented Generation (RAG), Cicero AI has significantly enhanced the quality and credibility of its chatbot. The result is a more trusted, efficient tool for legal professionals, automating time-consuming tasks like research and drafting. Cicero AI began the AICC engagement with limited knowledge of AI deployment and emerged with a fully operational LLM/RAG solution now live on the market. Thanks to comprehensive knowledge transfer, their team is now self-sufficient in evolving and managing the system, giving Cicero AI a distinct competitive edge in the legal tech space.

Image for Impact and Results

In Their Words

Christopher Williams, Co-Founder of Cicero AI

“The Transformer Programme helped us fast-track the development of domain-specific AI solutions that are truly aligned with the needs of the legal sector.

With the support of the AICC team, we were able to fine-tune our legal research tool, integrate jurisdiction-specific legislation, and significantly reduce manual effort for our clients.

The team understood our challenges right away and delivered impact quickly and remotely, which was essential for us. This collaboration has not only enhanced our product offering but also given us a real competitive edge. We’d absolutely recommend the programme to other innovation-driven businesses.”

Whats Next

Cicero AI plans to expand the fine-tuning and RAG setup across multiple areas of law, integrate user-specific datasets securely, and have already started to onboard early adopters. With a future roadmap centred on intelligent, regulation-aware tooling, they’re positioning themselves as a leader in providing ethical AI for legal services globally.

Please note: This case study was written by the AICC team, with support from AI tools including ChatGPT and Claude Sonnet 4 to assist with refinement.

Image for In Their Words

Glossary

We'd love to hear from you

Innovation thrives through connection. Whether you're an SME, researcher, or professional exploring AI, we’re here to help.

Our Partners