The $700 Billion AI Productivity Problem No One's Talking About

Market Urgency & Sentiment

• 85% of companies surveyed believe they have an 18-month window to become AI leaders or fall behind.
• 70% of enterprise leaders report believing they are currently wasting AI spend due to a lack of measurement systems.
• Despite narratives of AI "overhype," the speaker argues the technology is underhyped because productivity unlocks have not yet diffused broadly through organizations.

Measurement Challenges & Methodologies

• Traditional metrics like "amount of software bought" have become targets, rendering them inaccurate due to Goodhart's Law.
• Surveys are deemed unreliable for measuring productivity as respondents naturally bias answers positively to please employers.
• Laridon's approach combines "gold standard" productivity surveys with proprietary passive behavioral data (actual tool usage logs) to determine correlation between usage and output.
• A key emerging metric is "interdepartmental responsiveness" (e.g., faster responses from Legal to Product), used as a proxy for productivity without triggering gaming of the system.
• Passive measurement is preferred over self-reporting to avoid "principal-agent" problems where employees may inflate productivity feelings without actual output gains.

Adoption Friction & Employee Behavior

• Enterprise AI adoption is hindered by 80%+ of employees using shadow IT tools they are unaware are licensed by the company.
• Fear of looking "dumb" or violating security regulations prevents 40+ year-old employees from adopting new tools.
• Laridon's Engagement Strategy:
  • Wrapping LLMs to provide a "safe space" where employees cannot accidentally violate data policies or get fired.
  • Deploying customized models (e.g., Llama) that block illegal or company-forbidden prompts (e.g., HR data uploads in the EU).
  • Avoiding forced "top-down" training in favor of rewarding "hero" users who discover rapid efficiencies (e.g., reducing 8-hour tasks to 1 minute).
• Case Study: A 28-year-old investment banker created a 30-slide deck in one hour using ChatGPT; the speaker argues this individual should be made a "hero" rather than the sole subject of a corporate training deck.

Labor Market & Future of Work

• The speaker rejects the thesis of mass unemployment, citing competitive market forces: competitors will expand their workforce to outperform firms that cut heads.
• High-margin, solo-entrepreneur models (e.g., $1B revenue with one employee) exist but will not dominate the Fortune 500 due to competition.
• White-collar, hyper-educated workers face higher anxiety about displacement, though they possess the skills to retrain faster than low-skilled workers in previous industrial shifts.
• The transition is characterized by increased productivity expectations rather than reduced hours; employees must work "more" or "smarter" to maintain value.

Historical Parallels & Infrastructure

• The current AI cycle mirrors the 1990s internet boom, specifically the evolution of ad tech where infrastructure (Comscore, Nielsen, DoubleClick) was required to monetize and measure the new medium.
• Without measurement and governance infrastructure, the value of AI tools cannot be scaled or verified for enterprise budgets.
• Companies like Google and Facebook succeeded because the measurement stack existed; the same infrastructure is now needed for AI.

Vendor & Customer Dynamics

• AI vendors currently avoid third-party measurement to protect their "best-in-class" narratives; however, long-term adoption requires independent validation of value.
• Enterprise IT spend is projected to grow significantly (e.g., potential $18B to $180B for a giant like JPMorgan), requiring CFOs to shift focus from "adoption" to "value realization."
• "Cursor" is cited as an example where the tool transforms "mediocre engineers into good ones" and "amazing engineers into gods," creating a clear use case for measurement.

Product Marketing Insights

• Broad "horizontal platform" pitches (e.g., "we can do anything") are less effective than specific "tip calculator" use cases that solve immediate, high-value problems.
• Successful AI diffusion requires demonstrating specific ROI (e.g., specific market share data or tire safety analysis) rather than general capability.

Forward-Looking Statements & Strategic Intent

• Laridon aims to become the primary measurement and governance partner for AI, moving beyond just CIOs to include CMOs and CTOs.
• The company intends to replace traditional corporate health indices (McKinsey, Towers Watson) with real-time, behavioral data on AI productivity.
• The speaker predicts a future where "18 months" becomes the critical window for all enterprises to mature their AI governance.
• Closing Thought: Regardless of changing rankings (e.g., US News college rankings), certain institutions (like Harvard) retain their perceived status regardless of data fluctuations; similarly, the value of AI will be proven by long-term adoption, not hype cycles.