Rising With AI: Powering a New Era of Work

For the last event in our 2025 Beyond Silicon Valley Speaker Series, we headed across the Potomac to Amazon’s HQ2 for a conversation on AI and the future of work. What that future holds is a question looming over startups, sectors in transition, and workers everywhere, but the DMV has a unique vantage point, sitting close to the legacy systems AI is reshaping and closest to the policymakers trying to keep up.

Amazon’s Brian Kenner set the scene: AI is changing the talent market, and the D.C. metro area is feeling that shift earlier than most. With one of the highest per-capita concentrations of AI-native job postings in the country, the District is becoming an early test bed for AI’s impact on hiring, team structure, and sought-after skills.

A Turning Point for Workers and Workflows

Our panelists, Ardent Managing Partner Phil Bronner, Cooley Partner Adam Ruttenberg, and BuildWithin CEO Ximena Gates see the shift happening up close in their own work.

Bronner, who backs AI-native companies at the earliest stages, called this “the biggest transition I’ve been through, and I’ve been through the internet and mobile.” The demand for AI-fluency in job postings is up nearly 700%, and in his portfolio, teams are shipping meaningful updates every week as customer expectations accelerate.

Gates, who builds workforce infrastructure used by public- and private-sector employers, framed AI as a reset. “The clock started again.” For her, the DMV’s legacy systems offer real opportunities for modernization, and BuildWithin’s apprenticeship roots give her a practical lens on how workers adapt. The hill she’ll die on: companies that update their talent systems, not just tools, will win.

Ruttenberg, who advises companies on commercialization and data strategy, focused on the legal and operational volatility that comes with this shift. “The laws are squishy right now.” State rules diverge, policies evolve monthly, and teams need to build quickly while preparing to adjust as the ground shifts beneath them.

The New Toolbox for Building in the AI Era

Does becoming AI-native require a full reset? Not necessarily. As Bronner put it, “You build from where you are.” But there are a few non-negotiables:

• Real domain expertise paired with the technical ability to apply it
• Operations upgraded for AI before new products are layered on top
• Comfort working with unstructured data, multimodal inputs, and faster feedback loops
• A clear view of the customer and the problem they actually need solved

Gates noted that incremental fixes aren’t it. “Organizations don’t need just a point solution. They often need the whole guacamole.” Legacy systems have held companies hostage for years, and AI is exposing where deeper overhauls are overdue.

Ruttenberg added that building in this era requires practical judgment, from knowing how to commercialize and price AI tools to navigating the legal constraints that shape what’s in and out of bounds. “If it’s illegal for you to do it, it should be illegal for AI to do it. That’s the general principle lawmakers are trying to ladder up to.”

Bronner sees the strongest momentum from teams that can take 80% models and tune them to 100% based on the needs of their specific market. These are founders who iterate weekly, understand the nuances of their audience, and use that insight to shape how the model performs.

The Economics Behind the Acceleration

The conversation also surfaced real questions about cost, business models, and risk.

Ruttenberg noted that not every use case requires a large, expensive model, and many companies are turning to tools that balance capability with efficiency.

Gates sees pricing as just as much about communication as computation. “You cannot tell a customer something will cost a million tokens. You have to translate it into dollars and time.”

Bronner drew a line between the infrastructure race and the rest of the ecosystem. The bubble risk, he argues, sits with the model builders pouring billions into capacity, not with the companies applying AI to real markets. Thin wrappers will stay vulnerable, but the broader opportunity will still add trillions in net-new value.