Career Path: Your Secret Agent as Your Secret to Success

Banner artwork by Krot_Studio / Shutterstock.com

One of my favorite movies is Bridge of Spies with Tom Hanks. Hanks played James Donovan, a private practice lawyer who gets thrown into the Cold War world of spies, agents, and geopolitical turmoil. Donovan represented Rudolf Abel, a Soviet spy who was caught and charged with espionage. Donovan later negotiates the release of Gary Powers in exchange for the release of Abel. I love this movie because it shows how a lawyer can do anything – tackling challenges far beyond the law. Donovan was practicing insurance law one day and knee deep in espionage work the next! 

Turning to modern day practice of law, lawyers and agents are being thrown into the mix together once again but in a different kind of way. My team has been “exercising” together in experiment with Artificial Intelligence (AI) and learning with each exercise how AI is becoming a more frequent part of the practice of law. I wrote about how you should encourage the exercise approach to AI with your team, which you can read here. One of the keys to unlocking AI’s potential is the use of agents. AI agents are programs designed to interact with data and perform tasks autonomously to achieve specific goals as outlined by humans. 

Agents can be your secret weapon, augmenting your abilities and making you more efficient with your time. With each day, there is more information for lawyers to digest and understand to be effective in their practice of law. The practice of law in many areas has become so specialized that it takes one person consuming all the information to master one small corner of the one type of practice of law. Agents can give you a running start on an assignment, act as your project manager directing you to key developments to stay on top of your specialized area, or act as a thought partner on difficult issues. 

One of my team members, Bunmi Emenanjo, has really impressed me with her creation of agents and use of them in her practice of law. Bunmi is Senior Counsel for GSK’s vaccine business. The manufacturer of vaccines is a heavily regulated area and the job requires an enormous ingestion of data to be effective. I asked Emenanjo to answer some questions to help demystify the use of agents for in-house lawyers, including those lawyers who do not consider themselves tech savvy. Anyone can create and use agents as part of their practice of law – it is just a matter of investing the time that you will gain back later through your use of them. 

Bunmi Emenanjo, please first tell us a little about yourself and your role (including your tech skills) and how you got interested in the use of agents to further your practice of law? 

Bunmi Emenanjo: I’ve always being interested in science and technology. I have an undergraduate degree in cell biology and molecular genetics, earned a master’s degree in regulatory sciences from Johns Hopkins, and explored regulatory frameworks in synthetic biology and nanotechnology earlier in my career. In addition to the six years I spent at the FDA, I worked in the Office of General Counsel within the White House Office of Science and Technology Policy supporting the Artificial Intelligence Initiative and projects related to planetary protection. Emerging tech is my jam! Throughout my career, my focus has been helping innovative technologies reach the public in a legal and compliant manner. 

When I had the opportunity a few years ago to help develop artificial intelligence (AI) governance at my former employer, Otsuka Pharmaceuticals, I jumped at it. That experience developing the organization’s principles for the ethical and responsible use of generative AI laid the foundation for how I approach AI today. At the time, AI was regarded more tentatively; a lot has changed since then.  

In my current role as legal counsel supporting GSK’s vaccines business, I see AI as a collaborator that helps me work more efficiently and think more creatively about solving legal and business challenges. With the tools available to me at GSK and a foundational understanding of AI risks and how to mitigate them, my perspective has shifted from caution to a deeper exploration of AI’s possibilities. I truly enjoy exploring how I can use AI to optimize not only my work but also support my team.  

Emenanjo: Absolutely! No-code agents are increasingly easy and intuitive to set up. A no-code agent builder lets you create an AI agent without writing any programming code. Microsoft Copilot, Chat GPT, and Claude all offer versions of no-code agent builders. Think of agents as experts in specific subject matters — they draw on a defined universe of knowledge and complete tasks based on instructions and rules you provide. For example, a no-code agent can execute multi-step tasks such as analyzing voluminous documents, categorize the information within them, and create a summary without additional prompting from a human.    

Building a no-code agent in Microsoft Copilot using Agent Builder can be as simple as filling out a form that describes the agent’s role, includes instructions and operating rules, and points to a specific knowledge library. And here is the thing — AI can help you every step of the way, including drafting the instructions. You can tell the AI what you are trying to build, explain your objectives, provide relevant context, and ask it to help you create the agent. You can also ask it to explain what you do not understand or to simplify the steps.  

Emenanjo: I use agents to help analyze advertising and promotional questions where the answer is not obvious from one source alone. Because I have connected the agent to a vetted knowledge library that includes applicable laws, FDA regulations and guidance, clinical materials, and other relevant internal resources, the agent helps analyze the issue across the library, identify supporting sources, surface potential risks, and propose practical mitigation strategies. I still make the legal judgment, but the agent can give me a faster, more organized starting point. As a result, I can spend more time evaluating strategic legal considerations and less time gathering and organizing information. 

Emenanjo: When engaging with AI, it is important to treat it as a collaborator, not a decision maker. Human review of output is non-negotiable as inaccurate or incomplete output, including hallucinations, remains a risk. However, I have found that this risk can be mitigated by pointing the agent to vetted knowledge libraries and including specific rules within its instructions, such as requiring the agent to cite sources within the output so you can validate the work. AI can be persuasive even when its output is incomplete or incorrect. Resist the temptation to accept the output at face value; critical thinking and legal judgment remain essential. 

Another watchout is to ensure that you are inputting information into a secure, enterprise-approved platform. Be careful not to share proprietary, confidential, or privileged information on AI platforms that are not approved for those uses.  

Emenanjo: The most valuable aspect of my use of AI can be categorized into two areas — time savings and innovative thinking. The AI agents I built can ingest information spread across different formats, including applicable laws and regulations, clinical studies, and CDC recommendations. They can synthesize that information and analyze the issue holistically within minutes, a task that would typically take several hours. I have also found that AI agents challenge me to think through issues and risk mitigation in interesting ways. Because the Agent can ingest and hold voluminous information at once, it is able to review a fuller picture and suggest risk mitigation strategies that may not be immediately evident.  

Over the next two to three years, I can see AI agents becoming embedded in routine ways of working within legal departments. AI has the potential of becoming highly specialized legal collaborators that unburdens lawyers from administrative, simpler tasks allowing an intentional focus on high-value, strategic work that accelerates decision-making and drives business objectives. 

Emenanjo: I’d like an aspiring user of AI agents to walk away from this interview with these three tips: 

There are two elements of  Emenanjo’s use of agents that I think we can all learn from. The first is that she exhibits persistent curiosity. New technology can be daunting and lead us to “I will tackle it tomorrow” attitude when tomorrow never comes. Each day, you should set aside one task or project to test with AI and agents. Over time, you will be surprised at how quickly the use of AI becomes natural to you as part of your day. 

The second is that she is using AI to further her AI and agent knowledge and use. When she first thought of how to write instructions as part of a no-code agent, her first thought was to ask AI to help her write the instructions. She uses AI as a “thought partner” on improving her own use of technology. If you let AI be your secret agent, it can become your secret to success.