The Karnov Group, which is headquartered in Sweden and has bases in Denmark, Norway, France, Spain and Portugal, has launched KAILA (Karnov AI Legal Assistant), an LLM-based capability to take its extensive legal data collections to a new level of utility for lawyers.
KAILA delivers ‘precise and reliable answers along with source references, transforming legal workflows’. On top of improved legal research it uses genAI to help with:
- ‘Comparing legal provisions – Get a clear comparison of different legal provisions, e.g. by generating tables that highlight differences and similarities.
- Prepare legal arguments – You can gather arguments and practices that support your position in a legal case, e.g. to contest a termination or defend a trademark.
- Generate legal texts – Get help generating documents such as complaints, legal investigations, reports or other important legal documents adapted to your situation.
- Review and improve texts – Get help reviewing your legal texts and ask for improvements to make them clearer, more correct and in line with current practice.
- Analyze – Get an assessment of whether an action requires permission or notification under relevant legislation, e.g. within environmental law or company law.
- Ask KAILA to prepare you for meetings – Gather relevant legal information, potential complications and prepare questions to ask prior to meetings.
- Generate checklists and drafts – Get help creating checklists and drafts that ensure you comply with applicable regulations and best practices in your company.
- Summarize – KAILA can summarize documents for you, so you can get a quick overview of a case, law or agreement.’
In short, this is taking what LLMs can do, with their broad range of language and concept understanding, as well as text generation, and applying it to the huge Karnov legal library, using a chat interface.
One example they give is a lawyer typing in a prompt about a criminal matter: ‘I need inspiration for arguments that will secure my client the lowest sentence’, (see below).
KAILA first helps to rephrase the prompt, then taps into the statutes that apply and explains the legal issues around the charges to provide more context. Then you can dig into past cases that connect to this one, and then broaden out to other related case law, as well as the legislative history related to the matter. After that, based on the ‘skills’ set out above, you can then conduct a wide range of other tasks with the AI assistant, such as drafting.
The company’s key point is that by applying generative AI to their libraries of legal data you don’t just have better access to the information there, you have at your fingertips an assistant that can help generate answers to complex, sometimes open-ended questions in seconds, and then move into other tasks, all via one platform and which otherwise would have taken many hours to handle.
Or as the company explained: ‘KAILA not only finds the right sources quickly, but also assists with legal assessments and answers, paving the way for significant time savings in everyday work.’
Artificial Lawyer spoke to Head of Business Development, Selcuk Ünlü, about the new product. Here, he explains how they developed KAILA and what it means for their company and the users.
‘In the development of KAILA it was important for our AI team to understand how lawyers work and the needs they have. Therefore, we have had lawyers and developers working side by side, all the time challenging each other to make sure KAILA was built to meet their actual needs.
‘After beta testing since August and constantly adjusting according to the feedback we got, we are now ready to launch. In the test KAILA was praised for its ability to answer like a lawyer and for applying legal methodology. This is, of course an achievement, but I would argue that our biggest achievement lies in the Retrieval Engine and Intent Analyzer.
‘It’s crucial, for almost all legal tasks, that you need to work with the correct sources. If your starting point, your sources, is weak then that determines the quality of all steps that will follow, including your end delivery. KAILA is built to reason with the correct sources, sources that it has access to through our legal research databases.’
Impact
Is this a big deal? In short, yes. Karnov has a large library of legal information and by applying generative AI to this they have hugely broadened out what can be achieved. They have taken ‘static’ data and turned it into something that becomes actionable intelligence and delivers insights, along with the usual drafting, comparison and summarization tools one now expects with LLM approaches.
It’s all part of a wider movement around knowledge management and legal research where LLMs applied to data bases can create a major increase in value, while also reducing the time it takes to handle many legal tasks. We have seen other companies with large legal data bases seek to do the same thing, from Thomson Reuters, to LexisNexis, to vLex, and others, who all have seen that by adding a generative AI interface (with the right refinement, RAG and UI/UX tooling) to this information it really expands what can be done.
To conclude, now that genAI has arrived legal research will never be the same again. And it will be better for it…!