What I enjoy about AI enablement, and about what BD&P is doing in this area, is working through a real problem with someone. Where are they getting stuck? Could we make that part of their day easier? I find that exercise much more interesting, and more useful, than providing a product demonstration and leaving people to experiment on their own. 

Our AI Enablement Team is creating space for people to ask questions, even when they are not yet sure how to frame them. People often recognize friction before they can formulate a technical request. They should be able to describe the task, its context, and the desired outcome without also having to prescribe a solution. The important part is that the problem has somewhere to go. 

Consider looking for a specific point in a long transcript or contract. A keyword search may find the subject, but the relevant passage may express it in different language altogether. An approved AI tool can help identify candidate passages and provide references to check. Having access is not the same as knowing how to use the right tool for the right task. The question is how to connect what someone is trying to accomplish with the best workflow for the job. Sometimes the answer involves new technology. Sometimes a capability they have not discovered is already available in a tool their organization has purchased. 

Working alongside someone on that search task reveals the details: which information matters, what a good answer enables them to do next, and why an apparently simple request keeps failing. It also gives that person a chance to shape the solution. Someone who knows every awkward step in a process has knowledge a product demonstration cannot provide. This is where different kinds of expertise become valuable. The person who understands the work can explain why certain details matter. Someone familiar with the technology can suggest a different approach, test the idea, and help work through unexpected outcomes. Neither has to solve the problem alone. 

Learning the technology’s limits is a necessary counterpart to discovering its possibilities. AI can help with discovery, but verification remains a separate and necessary step. Using AI to conduct the search does not remove the need for judgment. If AI identifies a candidate passage in that transcript or contract, the person relying on it must still review the original document, including the surrounding context and any relevant qualifications. 

When an experiment works, the next question is whether it solved one person’s problem or a problem other people share. How many people across an organization face a similar task? If the answer is more than one, the useful part is not just the result. It is the method: what went in, which steps were followed, where judgment was required, and what limitations remain. Capturing that clearly lets someone else use what was learned, whether through an instruction, template, prompt, checklist, agent, or supported workflow. 

Not every experiment should become an organization-wide solution. A successful one should, however, be considered for reuse rather than left as an isolated result. That is the part of AI enablement I care about: giving a good idea somewhere to go, turning what works into something others can use, and creating value for the firm and better outcomes for our clients. 

Thank you to Joanne McKenzie-Hicks, BD&P Legal Research Specialist, for her assistance in preparing this article.