Tiffany Luck on AI Spending, Personal Agents, and Why Enterprise ROI Now Matters
Tiffany Luck on AI Spending, Personal Agents, and Why Enterprise ROI Now Matters
Tiffany Luck, a partner at New Enterprise Associates focused on early-stage AI, APIs, and B2B SaaS, joined TechCrunch's Equity podcast on June 17, 2026, to discuss the near-term outlook for AI company IPOs, the rising category of personal agents, and how enterprise buyers are shifting from "are we using AI?" to "what are we actually getting from it?"
Luck joined NEA in 2023 after working as a partner at GGV Capital. She sits on boards of several portfolio companies, giving her direct exposure to how AI spending decisions play out inside real organizations — not just in investor presentations.
The ROI Reckoning
Uber burned through its entire 2026 AI budget by April, consuming a full year of planned spending in four months, largely on Claude Code usage. By early June, the company had moved to cap individual employee AI spending. The COO openly acknowledged the difficulty of connecting rising token costs to improvements in customer-facing outcomes — a candid moment many engineering and finance leaders are grappling with privately.
Uber is not an isolated case. Meta built an internal tracking tool called "Claudeonomics" to monitor employee AI tool consumption, then shut it down in April 2026 after internal data began circulating publicly. The incident shows how sensitive internal AI spending metrics have become — organizations are simultaneously struggling to prove value and reluctant to discuss consumption patterns in public.
The broader context here is that enterprise buyers who moved fast in 2024 and early 2025 — often without formal measurement frameworks — are now facing budget reviews where the calculus has shifted. The question is no longer simply "are we using AI?" but "what returns did we actually get?"
From an investor's perspective, this inflection is not entirely unwelcome news. Startups that can credibly show productivity gains — more code shipped, tickets closed faster, deal cycles shortened — will have a meaningful edge in a market where buyers are asking harder questions. Luck's focus on APIs and B2B SaaS suggests she is backing companies positioned precisely in this measurement and instrumentation layer.
Personal Agents and the IPO Pipeline
Personal agents are transitioning from research project to product category. Earlier AI assistants were largely reactive — they answered questions or summarized documents when you asked. Personal agents operate differently: they act autonomously on behalf of users, handling multi-step workflows, retaining context across conversations, and managing authorization chains without human intervention at each step. The infrastructure demands are real — reliable tool integration, memory systems that persist and remain coherent, permission models enterprises will actually adopt, and response times that don't degrade the user experience.
For a venture investor with a portfolio in early-stage AI and API infrastructure, the agent layer is a natural point of focus. Agents are a forcing function for new API designs, new orchestration patterns, and new security and identity tooling. The question of how Luck views the near-term IPO pipeline for AI companies is worth watching. The 2025 market saw a handful of AI-adjacent public listings, but a meaningful cohort of pure-play AI software companies going public would be a significant signal of how public investors are pricing inference-dependent revenue at scale.
What Founders Should Know
Luck previewed her investment framework at TechCrunch's All Stage event in Boston in 2025, where she outlined five questions she expects founders to answer in pitch meetings. The Equity podcast appearance extends that thinking into the current environment — where distribution, measurement, and defensibility have become sharper as early generative AI enthusiasm has cooled.
The venture community is recalibrating more broadly. The wave of AI companies funded in 2022–2024 is reaching the stage where some must show sustainable unit economics rather than just rising usage numbers. Luck's focus on B2B SaaS and API infrastructure puts her in direct contact with this reckoning — she is backing the foundational infrastructure layer at the exact moment enterprises are asking whether the AI spending spree has actually paid off.
The Equity podcast episode is available via TechCrunch.
