The (Next) New Normal in AI and Generative AI
Since our founding, we have consistently emphasised two guiding principles for businesses adopting AI and generative AI. First, relying solely on what is already known is no longer sufficient. The pace of change in AI development means that our clients must prepare for shifts that may upend existing knowledge and strategies. Second, to truly future-proof a business, prioritising an integrated AI platform is essential. As AI becomes increasingly embedded in operations, firms need scalable, flexible infrastructures to keep pace with rapid advancements.
Recent developments in AI innovation and adoption have made these principles more critical than ever. The emergence of powerful yet cost-effective models, the democratisation of AI tools, and shifting geopolitical landscapes mean that businesses can no longer afford to take a passive approach. Instead, they must actively engage with these changes, adapting to remain competitive and resilient.
The Challenge of the “Unknowns”
In today’s fast-evolving AI landscape, mastering what is currently knowable is no longer enough. Business leaders must prepare for the unexpected—those “unknowns” that can disrupt industries overnight. Traditional approaches to technology, which rely on predictable roadmaps and steady updates, are woefully inadequate in this new reality.
Organisations must, therefore, cultivate adaptability as a core capability. Instead of following rigid implementation plans, they need to develop structures that allow them to pivot quickly in response to new advancements. This means investing in AI strategies that do not just solve today’s challenges but are equipped to evolve as new technologies emerge. The companies that thrive will be those that build AI ecosystems capable of integrating innovation seamlessly and efficiently.
Democratising AI Innovation
Historically, AI innovation was dominated by well-funded corporations, with only the largest technology firms able to afford the research and development necessary to push the boundaries of machine learning and computational reasoning. That era is now over. The emergence of open-source tools such as Sky-T1 has ushered in a new age of democratised AI, making sophisticated capabilities accessible to individuals and small teams.
Sky-T1, for example, is an advanced AI model designed for complex reasoning tasks such as mathematics and coding. It combines powerful performance with remarkably low costs, making it freely available to anyone willing to experiment. This democratisation of AI has radically altered the landscape of innovation.
The democratization of AI tools has created a new innovation ecosystem where breakthroughs can emerge from unexpected sources, such as start-ups or independent developers, often challenging established industry players. Open-source communities, which draw on diverse perspectives, can achieve remarkable results that frequently outpace those of traditional R&D teams. The conventional timelines for AI advancements are becoming increasingly unpredictable, making flexibility and responsiveness essential for organisations that wish to remain competitive. The next transformative AI tools may come from small, agile teams, or even an individual working with open-source resources.
DeepSeek R1: A Case Study in Disruption
The story of DeepSeek R1 illustrates the unpredictable nature of AI advancements. Developed by a Chinese start-up with a modest $6 million budget, DeepSeek R1 has disrupted the dominance of Silicon Valley’s AI giants. With exceptional performance, cost efficiency, and open-source accessibility, it has set a new benchmark for the industry.
Unlike traditional AI models that require enormous computing resources, DeepSeek R1 achieves remarkable outcomes through compact efficiency. It employs innovative learning techniques inspired by AlphaZero, which allow it to reach high-quality results without relying on conventional training methods. Additionally, its ability to explain its decision-making process introduces new levels of transparency and accountability, something that has long been a challenge in AI.
From an economic perspective, DeepSeek R1 is a game-changer. Its training costs were only a fraction of the $600 million spent on developing GPT-4, and its operational expenses remain significantly lower. This disruption has already had global ripple effects, including a measurable decline in the stock valuations of some US technology firms, demonstrating just how swiftly AI advancements can reshape markets.
The Impact on Harvey.AI and Legal-Specific Tools
These developments pose significant challenges for Harvey.AI, one of the most prominent AI tools designed specifically for law firms. While similar risks apply to other legal-specific tools, Harvey provides a useful case study in how rapid innovation in AI can disrupt even well-funded start-ups.
The rise of highly efficient models like DeepSeek R1 threatens Harvey’s business model, which relies heavily on OpenAI’s GPT. As newer, more cost-effective AI models become available, investor confidence in Harvey’s long-term viability is likely to wane. Unless Harvey adapts by integrating alternative technologies or reducing its dependency on OpenAI, it risks falling behind competitors who are quicker to embrace innovation.
If Harvey struggles to secure funding, the legal sector may experience a slowdown in the development of advanced tools tailored for legal professionals. Financial pressures could force Harvey to increase its subscription fees, making it less accessible to firms that have come to rely on it. More worryingly, if future funding rounds fail to provide sufficient capital, the company—still in its start-up phase—could fail altogether. In such a scenario, the legal sector may find itself dealing with inconsistent standards and fragmented AI adoption, a challenge it has faced in other areas of technology adoption.
The Risk of Reactive Technology Aversion
Disruptions like these may lead some law firms to take a cautious approach, using market volatility as a justification for avoiding AI adoption altogether. Such conservatism carries significant risks.
Modern clients expect their law firms to use technology to deliver more efficient and effective service. Law firms that fail to embrace AI may find themselves losing clients to more technologically advanced competitors. Additionally, the next generation of legal and ancillary professionals, who are often highly tech-savvy, may prefer to work at firms that actively engage with AI innovation, making it more difficult for traditional firms to attract and retain talent.
Rather than viewing AI adoption as an all-or-nothing decision, firms must reframe their approach. By implementing AI properly, they can balance innovation with risk management, ensuring they remain competitive without overcommitting to any single technology.
Navigating the New Geopolitical Landscape of AI
The rise of AI models like DeepSeek R1 highlights a broader shift in the global AI landscape. Where once the United States held a dominant position in AI development, new challengers from other regions, particularly China, are emerging. This shift introduces fresh considerations for businesses operating internationally.
Governments are increasingly linking AI development to national security, influencing how and where these technologies can be deployed. As a result, firms may need to be prepared to accommodate multiple AI and generative AI solutions, switching seamlessly from one to the other to meet diverse client expectations and regulatory requirements. Additionally, geopolitical tensions could impact access to certain tools, requiring businesses to remain agile in their AI strategies.
Lessons for Law Firms. The Case for Integrated AI Platforms
The rise of open-source models such as Sky-T1 and DeepSeek R1 underscores the limitations of single-purpose tools like Harvey. Law firms looking to future-proof their operations should consider adopting integrated AI platforms that offer scalability, flexibility, and a broad range of integrated capabilities.
Platforms like Palantir Foundry, for instance, enable firms to integrate multiple AI tools, allowing them to manage both routine and complex tasks with ease. By eliminating reliance on disconnected solutions, these platforms streamline operations, reducing inefficiencies and improving overall performance. Additionally, centralised data management ensures robust safeguards for sensitive client information, providing a level of data and cybersecurity that standalone AI tools struggle to match.
Beyond generative AI, integrated platforms incorporate analytics, workflow management, and other essential functions, offering a holistic approach to legal operations. In the case of Palantir Foundry, its intuitive design and rapid deployment capabilities provide a “time to value” that was previously unheard of in technology implementation.
The Bottom Line
The democratisation of AI and the rise of models like DeepSeek R1 highlight the rapid, unpredictable pace of technological change. For law firms and businesses alike, adopting an integrated AI platform is no longer a luxury—it is a business imperative. By prioritising flexibility, scalability, and a comprehensive approach to AI, firms can remain competitive in an era of unprecedented change.