Don’t allow your imagination to be captured solely by marketing campaigns. When selecting the AI vendor that is best suited for your business objectives, set specific goals and standards. Be clear what you want to achieve with the legal AI product and make this explicit: ask the AI vendor how their product could streamline your business processes and boost efficiency, not merely in abstracto, but in concreto. Don’t hesitate to ask critical questions as well.

‍

• How will this product streamline my current workflow?
• What types of output can the AI generate? Is it only capable of answering legal questions or can it provide contract drafts or write individual clauses as well? Can it draft other documents, such as a notice of default?
• Can the AI handle different types of legal research? Can it check for regulatory compliance as well as interpret specific contractual clauses? Is it easy to switch between the various features?

‍

Here’s an example. Say the bulk of your workload consists of contract reviews, a common requirement across many different sectors. In that case, you’ll need to find an AI vendor that understands the nuances of the specific tasks that contract reviews entail. You’ll need a legal AI solution that is designed to handle contract reviews in an efficient way. Anyone with contract review experience will understand that analysing a contract is not some abstract legal exercise. Reviews are always done from the perspective of a specific party. Clauses are drafted with the interests of a specific party in mind. Legal AI that is capable of performing general reviews that don’t consider the individual outlook of you or your clients will be too inaccurate to offer real value. The legal AI should be capable of integrating the outlook of the individual parties into its assessment of contractual provisions.

Developing legal AI is a team effort. It involves a very diverse skill set. A successful AI product requires decent software engineers, data scientists and AI specialists, but those people can only bring the product up to a certain level. If an AI solution is to have a real practical use in the legal field, the AI vendor should have a team of people with professional legal experience as well. If legal AI is supposed to think like a real lawyer, it should be developed with people who know what that thought process looks like, people who know which questions a lawyer deems important.

Furthermore, in order to develop a product that is up to industry standards, your AI vendor should engage in partnerships with industry leaders. The value of partnerships with prominent firms cannot be underestimated. It ensures a product that is practical and immediately deployable.

Additionally, if legal experts are involved, you will be able to provide detailed feedback about the product. Legal professionals can function as an intermediator between your company’s legal department and the AI experts developing the product.

Finally, expert legal guidance during the product development process will result in a product that is truly legal and not merely AI. Over the recent years, AI large language models (LLMs) have gained fame - and some notoriety - for their innovative potential, such as Chat GPT and Microsoft Copilot. While these product may be breaking new ground, they are not necessarily fit for legal applications. A distinction should be made between generic AI LLMs in and legal AI, because generic AI might produce outputs that are not reliable and lead to liability issues later on. In the long-term, using generic AI technology for specifically legal applications might prove to be an efficiency decrease instead of an increase.

‍

Here are some concrete tips:

• Mention the use cases you envision for the AI solution. Ask whether the AI is an appropriate means to be put to use in such a context. Has the AI vendor already tested the product in a similar context?
• Ask how you could provide feedback on the usefulness of the product. Will this feedback be used to improve the AI in the foreseeable future?
• Ask which partnerships with prominent firms your AI vendor has thus far established.
• Ask how your AI vendor has adapted and improved the LLM to be suitable for a legal context. What are the differences between this product and more generic LLMs?

Data is the backbone of AI. However, data poses two distinct problems.

Firstly, the training data should be qualitative. The quality - and quantity - of the data used to train the AI will have a decisive impact on the value of the output.

‍

In order to verify the training data quality of your AI vendor, here are some key questions:

• Which data does the AI model use for training? What is the source of this data? Is this data reliable?
• How often are the data refreshed? Do they remain up to date with new developments in the legal field?
• If the data sets are scraped, does the vendor have a license for using its data?

‍

The second problem posed by data is cyber security and privacy. In order to function properly, legal AI will need to have access to certain information: contracts, policies, terms and conditions, etcetera. It goes without saying that all of this data is of a sensitive nature.

‍

With regards to data security, these are some questions to consider:

• Will the vendor use your company’s data as part of its training data?
• What security standards does the AI vendor offer? Do they consider the legal compliance requirements to be sufficient or do they provide additional security warranties?
• How does the AI Vendor address security and privacy issues?
• What quality control procedures does the AI vendor adopt?

Implementing an AI-powered tool could bring radical efficiency improvements. However, AI is not magic, so you’ll have to consider how to concretely enhance your existing processes. This enhancement might entail some growth pains. It might even require your legal department to consider starting from scratch and devising an entirely new process.

In order to ensure that the change to your workflow is not too disruptive, you’ll want to choose an AI vendor that properly introduces your legal professionals to the features of the product. This calls for a thorough onboarding process and ongoing support during the integration process.

‍

In order to stay on top of change management, put these questions to your AI vendor:

• Is there a designated point of contact to address issues with the implementation of the AI solution?
• Is the AI vendor willing to schedule regular meetings during and after the initial implementation process?
• What about scalability? As your company grows and its needs evolve, will the AI solution be dynamic enough to continue to provide added value going forward?

AI can help increase efficiency and provide useful insights, but its limitations must be understood. Responsible use of AI, with human oversight, is important to mitigate potential risks in the legal domain.

‍

Here are some key questions:

• How likely is it that this AI will hallucinate? What steps has the AI vendor taken to prevent hallucinations?
• Is there a risk of incurring liability when using this product?
• Is the legal AI vendor currently facing litigation related to its product? Are there currently any regulatory inquiries into the company or product?
• Are there any possible privacy and security risks?

‍

While the price of the product is not really a risk, it does represent a cost that your company will have to bear in order to integrate the legal AI application. Therefore, as a final piece of advice, we would recommend to get a clear picture of what the cost of the product will entail.

‍

Here are some things you would want to consider:

• What is the price exactly and which services are included in the price?
• Are there any other costs associated with the implementation of the product?
• Does the vendor offer a flexible monthly subscription? Or do you need to commit for a longer period of time?

‍

‍

Conclusion: Ask Away

‍

When selecting an AI vendor, it's crucial to align the vendor's offerings with your business objectives. The best way to do this is by asking many, many questions. Key considerations include the vendor's technology, expertise, support services, and level of transparency. By carefully assessing these factors, you can minimize the risks associated with implementing the vendor's AI product while maximizing the return on their investment. This comprehensive approach ensures a balanced and informed decision that supports your business's long-term goals.