Interview about the use of (Gen) AI in the legal sector with Jonathan Voo (JSM)
This Interview is with Jonathan Voo, Head of Innovation at JSM
Could you give a small introduction of yourself and what you do at Johnson Stokes & Master (JSM)?
I'm Jonathan Voo, Head of Innovation at JSM, with a multidisciplinary background in computer science, law, and accountancy. My career in legal services extends across Europe and Asia, including previous experience as a Legal Project Manager overseeing global corporate transactions, reorganisations, and advisory matters. At JSM, my primary mandate is to future-proof our firm by evaluating and optimising our technological infrastructure, processes, and service delivery methods. This involves comprehensive assessment of our current legal tech toolkit and alignment with evolving client demands. A key focus area is the strategic implementation of generative AI, where we're exploring various applications including document summarisation, translation, data extraction, and intelligent review processes.
Our transaction group has successfully implemented various platforms to streamline due diligence processes and document automation. I'm particularly proud of our innovative client-facing solutions, such as a mobile application developed for a Global Non-Profit Organisation. This pro bono initiative transformed a 50-page PDF into an accessible, multilingual mobile interface, helping refugees and low-income individuals navigate Hong Kong's education system through intuitive charts and simplified processes. We're actively engaging with clients through AI workshops and employing design thinking methodologies to co-create solutions. This collaborative approach allows us to identify pain points and develop targeted improvements that benefit both our firm and our clients.
As one of the pioneering innovation leaders in Asia's legal sector, I focus on empowering our lawyers with cutting-edge tools while helping clients optimise their legal operations. My expertise lies in identifying and implementing transformative technologies that enhance operational efficiency and service delivery within dynamic, multicultural environments. I'm particularly skilled at harnessing disruptive technologies to drive organisational change and evolution in the legal sector.
Can you elaborate on how you use (Gen) AI in your daily operations at JSM?
At JSM, AI and legal tech have revolutionised our productivity, fundamentally transforming how we deliver legal services. We harness these technologies for routine tasks such as contract analysis and due diligence, enabling our lawyers to focus on strategic advisory work. Harvey AI, our primary tool, has markedly enhanced our efficiency by excelling in summarising lease agreements, identifying critical clauses, and streamlining translations from Mandarin or Cantonese. These capabilities allow us to accurately pinpoint important elements within documents—tasks that previously required several hours of a trainee or junior associate's time—while significantly reducing both costs and time investment.
Our communication standards are maintained through AI-powered grammar checking and email composition assistance, ensuring consistently professional communication that upholds our firm's reputation for excellence. We have used Gen AI for the creation of communication, moving beyond traditional PowerPoint or PDF presentations to create dynamic, interactive presentations that effectively engage clients and convey complex information with clarity. Our data analytics capabilities have been enhanced through generative AI implementation, enabling rapid decision-making, strategic planning, and performance analysis, proving invaluable in optimising our operations.
Whilst AI has dramatically improved our efficiency, we maintain rigorous quality control through a comprehensive checklist governing AI output verification, including multiple validation steps and detailed accuracy checks. This systematic approach ensures all AI-generated content meets our exacting standards before implementation. From document summarisation and translation to enhanced client communications and data-driven insights, these technologies have revolutionised our service delivery, whilst our commitment to excellence through careful oversight of AI outputs ensures we consistently meet and exceed our clients' expectations.
Apart from the benefits that you experience, did you also experience some drawbacks?
Regarding drawbacks, when generative AI first emerged, there was considerable uncertainty about its capabilities and limitations. During a cyber masterclass, I shared an incident where a lawyer used ChatGPT without proper verification, presenting AI-generated information in court that was subsequently challenged by the judge. I have seen some users treated it like a Google search engine, which highlighted the need for proper training and awareness.
One significant challenge has been client engagement. When working with clients' legal documents, obtaining consent for AI use initially proved time-consuming, particularly given the novelty of AI in the legal sector. Questions about data storage, security, and confidentiality needed careful consideration. Establishing comprehensive security policies and ensuring thorough understanding of the technology's advantages and limitations became crucial for our practice.
AI hallucination remains a significant risk that requires constant vigilance. We've invested heavily in training our lawyers to write more effective prompts, as we've learned that prompt specificity directly correlates with output quality. To facilitate adoption, we've developed a prompt library tailored to different practice groups, providing structured starting points for AI implementation. Additionally, we face ongoing discussions about AI-related cost structures. Clients naturally expect reduced fees when AI is employed, given the increased efficiency. However, these tools represent significant investments, and we must carefully balance efficiency gains with the need to demonstrate value to our clients whilst maintaining profitable operations.
What ethical dilemmas arise when you use AI technologies, how do you navigate these?
I mean, there is bias in training data and opaque AI decisions, pose real challenges. At JSM, we audit tools regularly, we involve lawyers, verify output and enforce strict data protocols. We make sure that we know where our data is, and that people use it correctly. We have dashboards let us know how much the tools are used, because at the end of the day, we need to make sure that people are using it and that is not stagnating. We also have training programs to ensure teams understand ethical AI, balancing innovation, and accountability. As I mentioned before, AI can sometimes produce biased results or hallucinated information, so it is crucial to make sure that our employees use the tools correctly and that the data is checked frequently. This also poses an ethical challenge, because we want our lawyers to feel comfortable and to reassure them that their work remains private. We only look at the use of AI a very high level – for example tracking the number of prompts that are used per day. Additionally, we receive a lot of feedback from our lawyers to understand how they are using it and what improvements can be made.
You already mentioned the issue with privacy and data storage. How do you tackle these issues?
Before implementing any new tools, we follow a rigorous procurement process through our IT security and risk management teams. We meticulously gather detailed information from Microsoft and other vendors to map the precise location of data storage. Maintaining data on secure Cloud systems and establishing robust ring-fencing ensures that sensitive information remains within the law firm's secure environment. This level of security is critical, as a data breach or third-party compromise could have severe consequences for our firm. We employ multiple layers of security, including sophisticated firewalls, comprehensive privacy measures, and stringent data protection protocols. Our data management includes automatic purging processes - for instance, with Harvey, all information is permanently deleted after seven days. We also utilise advanced redaction tools to remove names, addresses, and other sensitive client information before uploading any documentation to these platforms, ensuring complete anonymity and traceability protection. Furthermore, we maintain vigilant oversight of international regulations and governance requirements across different jurisdictions. This ensures our continuous compliance with evolving legal frameworks and data protection standards worldwide. Our approach to data security isn't static; we constantly review and update our protocols to align with emerging best practices and regulatory requirements.
You mentioned the fast developments in AI and its regulatory framework. How do you manage this as innovation manager?
Maintaining current knowledge of global regulations presents a significant challenge. Historically, we relied on a team of assistants to conduct research across our numerous jurisdictions, including China, Pakistan, Hong Kong, Singapore, the Philippines, Australia, Taiwan, and South Korea. Their methods included using monitoring news feeds, and even tracking LinkedIn updates to stay abreast of regulatory changes.
However, manually sifting through thousands of websites daily proved inefficient. In response, I'm currently developing a regulatory monitoring tool that harnesses generative AI to streamline this process. The tool automatically scrapes news feeds, articles, and regulatory updates from across the globe, then synthesises this information into concise bullet points. This technological solution significantly reduces the time and resources previously required for comprehensive regulatory monitoring.
To complement this digital approach, I actively participate in industry seminars and forums led by thought leaders in the field. Networking with fellow legal professionals and closely monitoring government websites provides valuable insights and helps ensure we maintain a comprehensive understanding of the regulatory landscape. This combined approach of technological innovation and traditional professional networking enables us to stay effectively informed of regulatory developments across all our operational jurisdictions.
What would be your advice to managers on how to make sure that AI tools are implemented correctly?
For managers implementing AI for the first time, I recommend beginning with establishing and communicating your strategic approach through a comprehensive AI policy. Collaborate with your legal teams, IT departments, and security specialists to establish proper frameworks before rolling out to team members. Understanding your objectives and identifying the right fit for your organisation is crucial. While many companies gravitate towards Microsoft's solutions, numerous capable LLMs and generative AI tools exist in the market. Conduct pilot schemes with four or five selected platforms, engaging power users to thoroughly test these systems before making a final decision.
Training employees in effective AI prompting is paramount. Set realistic expectations about results and encourage regular usage, as proficiency develops through consistent practice. This isn't something that can be mastered through passive learning like watching YouTube tutorials; it requires hands-on experience and continuous refinement. Emphasise the importance of enjoying the learning process and maintaining an experimental mindset.
Finally, ensure regular team check-ins, improving guidance, and updating regulations. The legal landscape surrounding AI is rapidly evolving, as is the technology itself. While generative AI dominated discussions last year, autonomous agents have become the focal point in 2025. Maintaining currency with developments, monitoring system usage, and soliciting consistent feedback from users helps ensure optimal implementation and adaptation of these technologies.
Do you struggle with motivating employees to use AI?
Initially, there was some apprehension and uncertainty about AI adoption, but demonstrating its capabilities led to positive uptake. At JSM, we implemented various tools including automation systems, document review platforms, mobile app development solutions, and no-code applications. The latter enabled lawyers and other staff to build applications independently. While these tools were already in place, the emergence of ChatGPT in 2022 catalysed unprecedented interest in AI technology.
ChatGPT's global presence and user-friendly interface made it one of our most successfully adopted tools. Its relatability and practical utility resonated strongly with our team members. Though the initial phase required time for people to understand the associated challenges, conducting demonstrations and explaining the ethics, reasoning, benefits, and limitations proved instrumental in gaining widespread acceptance.
We acknowledge that some individuals prefer not to adopt these technologies, which we respect. Our strategy focuses on engaging enthusiastic adopters who become natural ambassadors for the technology. We particularly target junior lawyers and young partners as pioneers, recognising their potential to influence more hesitant colleagues. This approach has created a ripple effect, gradually encouraging broader adoption across the firm through peer influence and practical demonstration of benefits.
How would you say that Asia, specifically APEC region, is positioned in adopting AI for the legal industry?
Our data dashboards provide comprehensive analytics on global user engagement, allowing us to track regular usage patterns, identify power users, and analyse implementation methods across different regions. Whilst our American offices currently lead in adoption rates, Asia holds a strong second position in our global rankings.
In the Asian market, Singapore, Hong Kong, and China are particularly notable for their AI engagement. China's technological advancement is evident through the development of sophisticated tools like DeepSeek. Hong Kong's dynamic startup ecosystem frequently initiates collaboration opportunities, reaching out to explore and understand various AI implementations. This consistent market activity and demand indicates sustained momentum in Asia's AI adoption journey. While these Asian markets demonstrate strong growth, other JSM offices worldwide are rapidly advancing their AI implementation efforts, gradually closing the adoption gap.
You have mentioned DeepSeek already, did your perspective change on AI after the launch?
DeepSeek distinguishes itself from ChatGPT primarily through its competitive pricing structure and comparable accuracy in output. The lower cost point makes it more accessible for widespread workplace implementation. While DeepSeek's Chinese origin raises certain considerations regarding data management and storage protocols, similar concerns exist with ChatGPT and other platforms. Ultimately, the choice between providers largely depends on your organisation's preferences regarding data governance.
DeepSeek's competitive advantage lies in its cost-effective operation and ability to provide more detailed, granular information. Although I haven't fully explored its capabilities, it's evident that DeepSeek is rapidly advancing in the LLM space. This development suggests a trend toward more accessible AI tools, potentially democratising access to advanced language models across different market segments. These emerging alternatives to established platforms like ChatGPT indicate a maturing market where competition drives both innovation and accessibility. The entry of more cost-effective solutions could accelerate AI adoption across various sectors, particularly in markets where cost has been a significant barrier to entry.
What will be the impact of AI in 5-10 years?
The legal profession's future remains secure, though significant adaptations are inevitable as technology evolves. Modern lawyers must expand beyond traditional legal expertise to embrace legal technology tools. Universities are likely to incorporate more AI-focused content into their curricula, and younger lawyers will face increased expectations to demonstrate both legal and technological proficiency.
A critical challenge facing the industry involves the economics of AI implementation - specifically, how to structure charging for AI usage while demonstrating value to clients. This complex issue remains unresolved, with many law firms actively seeking solutions. As more tools emerge, clients will simultaneously push for cost reductions, creating pressure for law firms to deliver services more efficiently and economically.
For firms like JSM, maintaining a competitive edge requires staying ahead of technological advancements. The stakes are clear: adaptation is crucial for survival in the evolving legal landscape. As one trainee aptly noted, "AI won't replace lawyers, but you will be replaced by someone who knows how to use AI." This encapsulates the transformation occurring within legal practice, where technological competency becomes as essential as legal expertise.
How do you use AI in your daily life?
Beyond the professional sphere, my enthusiasm for technology extends into daily life, particularly through home automation. My Google Assistant manages various household functions - from lighting and climate control to music - all triggered automatically when my phone connects to the Wi-Fi network. In addition, AI has become valuable in supporting my son's Chinese language education. I leverage AI technology to analyse photographs of his homework pages and provide appropriate responses. For last Christmas, I experimented with AI-generated cartoons and storytelling to create unique children's books as gifts.