Tiffany Luck: Vertical AI, an Asset for Startups

Tiffany Luck and Her Journey to Vertical AI

Tiffany Luck, a partner at New Enterprise Associates (NEA), has a rich background that has led her to focus on investing in the application layer of artificial intelligence (AI) and B2B SaaS. Before joining NEA about three years ago, she worked at early-stage startups like Lot18 and played a key role in developing consumer packaged goods (CPG) e-commerce at Amazon, long before the company acquired Whole Foods. She also gained valuable experience in technology mergers and acquisitions at Morgan Stanley and was a partner at GGV Capital, now known as Notable.

Today, Tiffany Luck is concentrating her efforts on vertical AI, seeking to bridge the gap between horizontal models and tangible return on investment (ROI) for businesses. She is betting on the potential of AI to transform the "last mile" of automation, where companies can truly leverage the technology.

The Evolution of Commerce and AI Adoption

In a discussion with Crunchbase News, Tiffany Luck spoke about the evolution of commerce, from the early days of Amazon Fresh to the current rise of AI. She sees parallels between the early days of e-commerce and the adoption of AI today. Back then, she had to convince CPG manufacturers that online commerce was the future, despite considerable technological and mental resistance. Today, Fortune 500 companies face a similar situation with AI.

While the potential of AI is evident, many organizations struggle to integrate it into their daily processes. AI must transition from a mere "shiny object" to a solution capable of solving concrete problems, which requires overcoming significant initial resistance.

Building Sustainable Defenses in AI

Tiffany Luck also addressed the "Anthropic question," a concern regarding the potential dominance of cutting-edge models over the application layer. Vertical startups can, however, defend themselves by focusing on the "last mile" of automation. Horizontal tools, such as Claude, currently serve as research copilots but fail to manage crucial final tasks.

To create solid barriers, startups must tackle the specific challenges of this last mile. For instance, in financial planning and analysis, it is not enough to integrate data into a general model. Startups need to be able to automatically forecast, signal specific trade-offs, and create a unified data layer from disparate sources. Startups that build these custom-designed product flywheels utilize engineers deployed upfront to work alongside users and identify gaps in workflows, thereby creating a competitive advantage that is difficult to replicate.

The Value of Workflow Ownership

Owning the end-to-end workflow is becoming increasingly valuable, as it eliminates the mental friction associated with using AI. If a company can deliver a finished product that meets the specific needs of its clients, the return on investment is undeniable.

Companies like August, which specialize in legal due diligence, and Samaya AI, which focuses on equity research reports, illustrate this point. When the end result is a document that meets or exceeds expectations, the company cares less about the underlying model and more about time savings and the accuracy of the results.

Partnerships and Competition with Platforms

Tiffany Luck also discussed the evolution of operating systems and how startups should consider partnering with or competing against platforms like Claude or OpenAI. While user interfaces have not yet radically changed, she anticipates a future where a model could become the default operating system, directly integrating specialized applications.

Just as startups used Slack as their main interface a few years ago, one could envision a future where specialized tools like Samaya are integrated into the user interface of a horizontal model. The specialized knowledge graph and proprietary data remain with the startup, but execution takes place within the user's main "operating system." Interoperability between these applications will be crucial for the future of operating systems.

Decisive Factors for Enterprise Buyers

In regulated industries, decisive factors for enterprise buyers include accuracy, auditability, and cybersecurity. Companies are concerned about the provenance of data and must be able to audit every figure.

The AIUC (Artificial Intelligence Underwriting Co.) is working to establish a certification standard for AI agents, similar to Moody's for financial ratings. AIUC has gathered a group of over 100 CISOs to create a for-profit certification standard. This standard aims to provide an additional layer of trust beyond existing standards like SOC 2 for companies like ElevenLabs or Cursor.

The Next Frontier for Investors

For Tiffany Luck, the next frontier for investors lies in the "pre-mobile native" era. While we have already shifted the web to mobile, truly innovative applications enabled solely by AI have yet to emerge. She is awaiting a "Waymo moment," where workflows will become truly autonomous and agentic, fundamentally transforming the way we work.

The next 12 months will be crucial in revealing these new ways of working, which could be radically different from the current era dominated by laptops and keyboards.