Managing Partner at Union Square Ventures (USV)
Check size: USV publishes a firm-wide check range of $1 million to $30 million and invests from the earliest stages through follow-ons. It does not publish Rebecca's individual authority, and the partnership makes decisions collectively without a vote.
Rebecca invests at the edge of large markets, with a consumer and product lens sharpened by USV's network-effects thesis. Her current view links abundant AI intelligence, cheaper hardware, sensors, and edge processing: newly reachable data from bodies, infrastructure, environments, and machines can power self-improving products and physical systems. She also expects open agent infrastructure to enable a wave of useful, creative, networked consumer applications outside dominant model platforms.
Open with the user or physical-world problem and the newly available capability. Show what data was previously dark, why sensing and inference are now economical, how deployment generates proprietary learning, and what action the product can take. For consumer AI, demonstrate delight and usefulness, model independence, alignment, retention, and the participation loop that can become a network effect.
Previously inaccessible data that becomes collectable and actionable, products that create a deployment-data-model-improvement loop, applications made possible rather than merely more efficient by AI, strong consumer insight, open and aligned agent infrastructure, fun and unusual products, fresh founder hustle, and experiences where each user or action strengthens the network.
Her published view distinguishes new capabilities from efficiency alone. It implies weaker fit for generic AI wrappers, products captive to one model without differentiation, physical systems lacking a credible data flywheel, consumer apps without sustained utility or network effects, and incumbency-minded products that do not exploit newly accessible data. These are inferred filters.
Identify a high-value dataset that was too costly, messy, or inaccessible to use and show which sensing, hardware, model, and product changes now make collection and action economical.
Measure how each deployment produces real-world data, improves the model, lowers the cost or time of the next deployment, and compounds a defensible advantage.
Separate products that make an old task cheaper from those that enable insight or action that could not previously be achieved at all.
Show how a useful or delightful agent action attracts use and how users, shared artifacts, data, or transactions improve the experience for subsequent participants.
As intelligence becomes ubiquitous, durable product experience, positioning, narrative, and founder-level product judgment become more important rather than less.
Cheaper sensing, hardware, and intelligence make previously inaccessible physical and biological data actionable, creating deployment-data-model flywheels across robotics, infrastructure, health, industry, and energy.
Model competition and open agent infrastructure create a window for weird, useful, creative consumer applications, especially where agent actions produce network effects.
+ 207 more investments. View fund →