The AI investment landscape is evolving at breakneck speed, creating both unprecedented opportunities and complex challenges for venture capitalists. At HubSpot for Startups' annual AiSummit in San Francisco, we brought together five leading AI investors to discuss where the smart money is going and what they're walking away from in today's market.

Our panel featured Adam Coccari, Managing Director of HubSpot Ventures, moderating a discussion with Adina Tecklu from Khosla Ventures, Cathy Gao from Sapphire Ventures, Priya Saiprasad from Touring Capital, and Rob Toews from Radical Ventures. These investors collectively represent some of the most active firms in AI funding today, offering insights into how they're navigating one of the most exciting yet confusing times in venture capital.

Identifying Overhyped AI Categories

The conversation kicked off with a reality check on AI hype. Despite the transformative potential of AI across industries, our panelists identified several categories where expectations may be outpacing reality.

AI for Coding and Software Development topped Rob Toews' list of overhyped areas. While acknowledging it as one of the most important applications for large language models today, he pointed out critical concerns:

• Unclear durability of competitive positioning
• Intense focus from big tech players and major labs
• Extremely lofty valuations due to market froth
• Questions about long-term stickiness of these products

Priya Saiprasad echoed these concerns while highlighting another overhyped category: end-to-end agentic workflows promising minimal human supervision. The reality in enterprise sales reveals significant challenges:

• Disparate and broken enterprise solutions still require human integration
• Consultancy and professional services remain heavily involved
• Fully automated, end-to-end promises often fall short of reality

Adina Tecklu identified a class of companies that "feel much more like features than durable products" — particularly note-taking apps and transcription services. While these provide utility and show revenue growth today, the critical question remains whether they'll exist as standalone companies in five to ten years.

The Great Shift: From Infrastructure to Applications

A notable trend emerged from the discussion: the marked shift in VC interest from AI infrastructure and foundation models to the application layer. Cathy Gao provided concrete data from Sapphire Ventures, sharing that of roughly 13 new deals last year, the vast majority were application companies rather than infrastructure or dev tool companies.

The reasoning behind this shift centers on differentiation opportunities:

• Application layer advantages: Companies can tap into industry-specific datasets, understand particular workflows, and separate themselves from broader competition
• Model layer consolidation: Fewer models rising to the top with declining prices
• Infrastructure maturity: Lower layers now support a Cambrian explosion of applications

However, Rob Toews offered a nuanced perspective, emphasizing that the relationship between infrastructure and applications isn't simply sequential. Infrastructure advancements continue enabling new application opportunities, while racing applications create needs for new model-level innovations.

Particularly exciting infrastructure opportunities exist in:

• New data modalities: Video, 3D, robotics, biology, music, audio, and tabular data foundation models
• Agent infrastructure: Payment systems, authentication, and orchestration tools for multi-agent systems
• Specialized tooling: Developer tools for the next generation of AI applications

Vertical vs. Horizontal: The Case for Industry-Specific AI

The panel strongly favored vertical, industry-specific AI solutions over horizontal approaches. Priya Saiprasad made a compelling case for this direction, citing the changed competitive landscape:

Why Verticals Win:

• Large hyperscalers can quickly close feature gaps in horizontal solutions
• Vertical markets offer opportunities to build data moats and network effects
• Industry-specific solutions provide protection from broad-based competition
• Real ROI delivery in targeted use cases creates stronger defensibility

Framework for Vertical Investment:

• Large enough TAM to support a standalone business
• Acquisitive industry players with legacy technology seeking partnerships
• Strong buyer propensity and willingness to pay for solutions

Real-World Success: The Elise AI Case Study

Cathy Gao shared a detailed example of vertical AI success with EliseAI, a conversational AI platform serving the residential real estate industry. The company's journey illustrates key principles of successful vertical AI:

Initial Problem Identification:

• Poor customer experience in apartment hunting and tenant services
• Limited office hours creating business losses
• Repetitive inquiries falling within predictable patterns

Strategic Execution:

• Deep integration into property management systems
• Understanding of customer workflows and industry regulations
• Quick adoption of new AI technologies as they became available
• Expansion from text to multimodal communications (call, text, email)

Results and Expansion:

• Over $100 million ARR today
• 90% automation of customer inquiries
• Successful expansion into healthcare
• Strategic partnerships with industry leaders like Zillow

Evaluating Founders in the AI Era

The rapid pace of AI development has fundamentally changed how VCs evaluate founding teams. Traditional metrics remain important, but new factors have emerged:

Critical Founder Traits:

• Agility and flexibility: Willingness to rebuild products as models evolve
• Vision scope: Ambitious thinking about future possibilities
• Decision-making: Ability to act with limited information
• Learning mindset: Capacity to adapt to constant technological change

Adina Tecklu highlighted Scott Wu from Cognition as an exemplary founder who embodies these traits, particularly his willingness to "throw out the product and start from scratch" as models continue evolving.

The panel also noted a fascinating trend: companies reinventing themselves rapidly, sometimes making full pivots or moving into adjacent areas within months. This has led investors to focus even more heavily on team quality over specific metrics or business models.

Industries Primed for AI Transformation

When asked about which industries will see the biggest impact from vertical AI agents in the next two to three years, the panel highlighted several key areas:

Scientific Applications emerged as a particularly exciting category. Rob Toews described the potential for AI agents to carry out the full scientific process:

• Reviewing relevant literature
• Generating novel hypotheses
• Creating plans to test hypotheses
• Potentially conducting experiments (with human or robotic assistance)
• Interpreting data and generating conclusions

This paradigm is already emerging in biology, therapeutics, material science, and AI research itself.

Framework-Based Approach was emphasized by Priya Saiprasad, who stressed the importance of evaluating verticals based on:

• Sufficient TAM for standalone businesses
• Acquisitive industry players with legacy systems
• Strong buyer propensity and willingness to pay

The Pricing Revolution: From Seats to Outcomes

The discussion revealed a fundamental shift happening in AI business models. Traditional seat-based SaaS pricing faces a critical challenge: successful AI products reduce the need for human employees, creating natural contraction rather than expansion.

Emerging Pricing Models:

• Consumption-based: Pay for usage rather than seats
• Outcome-based: Revenue sharing tied to actual results delivered
• Hybrid approaches: Minimum commitments plus variable pricing

Priya Saiprasad shared the example of Pyxis.ai, which helps generate return on ad spend and takes a percentage of the ROI they create. This aligns startup incentives with customer success while creating natural growth opportunities.

The challenge remains achieving the predictability that both VCs and companies need for business planning, especially when dealing with seasonal or cyclical business models.

Building AI-Native Culture

Modern AI startups are implementing new organizational strategies to maintain their competitive edge:

Internal AI Adoption as a Hiring Signal:

• VCs now ask about companies' internal AI stacks during diligence
• Not using coding agents or productivity tools becomes a red flag
• AI budget allocation is often unlimited for productivity tools

Hiring Strategies:

• Explicit preference for hiring ex-founders who bring agency and multifunctionality
• Lean teams supported by comprehensive AI tooling
• Companies scaling to $60 million ARR with just 30 employees
• Cross-functional teams operating like mini-startups within startups

Cultural Elements:

• Company-wide expectation of AI adoption across all functions
• Regular education and hackathon programming
• Top-down prioritization of AI efficiency

Investment Reflections: Learning from Success and Mistakes

The panel concluded with honest reflections on recent investments. Cathy Gao shared the success story of Moveworks, which seemed overpriced in 2019 but was recently acquired by ServiceNow for nearly $3 billion. The lesson: sometimes what appears expensive in the moment proves prescient when fundamental technology shifts accelerate company value.

Conversely, several panelists acknowledged investments in sales enablement and automation tools that struggled as AI changed the fundamental economics of sales team productivity.

Rob Toews highlighted Datology as an investment that initially seemed risky when everyone expected to rely solely on OpenAI or Anthropic APIs. However, the trend toward companies building and owning their own models, accelerated by developments like DeepSeek, has created tremendous tailwinds for data optimization platforms.

Key Takeaways for Founders

The panel's insights offer several actionable guidelines for AI startup founders:

1. Choose vertical specialization over horizontal approaches for stronger defensibility
2. Focus on engagement over revenue in early stages to build customer entrenchment
3. Embrace pricing model innovation aligned with the value you deliver
4. Build agile, learning-focused cultures that can adapt to rapid AI advancement
5. Prioritize outcome-based value propositions rather than feature comparisons
6. Invest heavily in internal AI adoption to maintain competitive advantages

The AI investment landscape continues evolving rapidly, but the fundamentals remain: identify real problems, build defensible solutions, and maintain the agility to adapt as the technology transforms entire industries. As these leading investors demonstrated, success in AI investing requires both technical sophistication and timeless business judgment.

AI Disclaimer: AI helped us summarize the key points from these videos, and our editorial team reviewed everything to make sure it's clear and correct.

Why wait for months to get access to all the videos when you can get it now? Sign up below and not only will you get early access to the full set of video sessions from this year's summit, but you'll have access to the videos from the past two summits as well; which alone is over ten hours of educational content and insights! Fill out the form to gain instant access to hours of AI video goodness!