NEA’s Tiffany Luck on AI’s ROI reckoning and the future of personal agents

The era of unchecked AI spending in Silicon Valley is giving way to a more sobering question: where is the return? On a recent episode of Bitcoin World’s Equity podcast, NEA partner Tiffany Luck sat down with senior reporter Rebecca Bellan to discuss the transition from the tokenmaxxing trend — where companies pushed AI usage to its limits — to a period of intense ROI scrutiny.

From tokenmaxxing to accountability

Earlier this year, tokenmaxxing was the dominant trend in Silicon Valley, with CEOs encouraging employees to maximize AI tool usage. But the enthusiasm has cooled as companies confront real costs. Uber reportedly burned through its annual AI budget in a matter of months. Some organizations have cut back on Claude licenses for parts of their workforce. Meta quietly shut down its internal AI usage leaderboard.

Luck, whose career began convincing companies that e-commerce was the future, now focuses on where value is actually being created in the AI stack. She noted that the shift from hype to ROI is creating a new set of challenges and opportunities for startups and enterprises alike.

How enterprises are measuring AI spend

Startups are stepping in to help companies track the return on their AI investments. According to Luck, forward deployed engineers are becoming a ‘Trojan horse’ for AI adoption, embedding directly within client organizations to demonstrate practical value. Rather than committing to a single model provider, many enterprises are now mixing and matching models depending on the task.

Luck emphasized that value is being created at every layer of the AI stack, not just at the model layer. This includes infrastructure, deployment tools, and specialized applications that solve real business problems.

The rise of personal agents

One area Luck is particularly optimistic about is consumer-facing AI, specifically personal agents. She described the potential for ‘magic moments’ where AI assistants anticipate user needs seamlessly. While the technology is still evolving, she sees significant opportunity for startups that can deliver reliable, personalized AI experiences.

What this year’s AI IPOs signal

The conversation also touched on the wave of AI companies going public in 2026. Luck offered a measured perspective, noting that while the IPO market is active, investors are demanding clearer paths to profitability than in previous tech cycles. The days of funding AI companies on promise alone are fading.

Conclusion

The tension between AI hype and measurable business outcomes is defining the current moment in technology. As NEA’s Tiffany Luck outlined, the companies that will thrive are those that can demonstrate real-world value, whether through personal agents, enterprise tools, or infrastructure. The tokenmaxxing era may be over, but the foundation for sustainable AI adoption is being built now.

FAQs

Q1: What is tokenmaxxing?
Tokenmaxxing was a Silicon Valley trend earlier in 2026 where companies encouraged employees to maximize usage of AI tools, often without regard for cost or ROI. The trend has since declined as organizations face budget overruns.

Q2: How are enterprises measuring AI ROI now?
Many companies are using specialized startups and forward deployed engineers to track AI spending and outcomes. Enterprises are also mixing multiple AI models rather than relying on a single provider, and demanding clearer metrics before scaling usage.

Q3: What does Tiffany Luck see as the biggest opportunity in AI?
Luck is particularly focused on consumer-facing personal agents and the broader AI stack beyond just models. She believes value is being created across infrastructure, deployment, and application layers.