The more useful AI gets, the more dangerous it becomes to use casually. That tension is the wedge Venice is driving into the market, and the route it’s taking to become a mainstream rival to OpenAI and Claude.

That may sound like a lofty ambition, but consider the differentiators: private by default, uncensored, and built to serve up the best model for a job, rather than lock users into one lab’s worldview. Together they make for a safer, cheaper, and more permissive AI experience, which is what we all want, crypto or not.

Jesse Proudman, Venice’s CTO, and Jon, its head of strategy, came on Bankless to lay out how the company turns those advantages into its ambition: win consumers with private, uncensored, aggregated AI, then win agents by selling that same inference to the machines running the economy underneath.

The Consumer Front

You’ve probably typed something into ChatGPT you wouldn’t want published, so you already understand Venice’s pitch.

AI has become the place people dump everything: health records, legal questions, business plans, the stuff they’d never post anywhere. It all sits in the largest databases of intimate human thought ever built, one subpoena, breach, or policy change away from exposure. Companies move the goalposts once they’re sitting on a trove they can turn into a competitive edge, and in a race this cutthroat, assume the majors are already mining your data.

Venice has no such reason, because it doesn’t build a model of its own. It routes to whichever model you want, or sends you to the best one for the task, frontier or open-source, meaning it has no incentive to hoard your data to compete. The same design answers the daily flood of new models, each better at something, where backing one lab means missing what the rest do best. A router has nothing of its own to defend, so no reason to police your prompts either.

Its incentives and the user’s point the same way.

The Inference Front

AI’s center of gravity is shifting from training to inference, and crypto is pricing that shift through the “Inference Capital Markets” narrative. I wrote about how Venice sits at the center of it, and Jesse and Jon confirmed it’s core to their strategy.

The driver is agents. A person firing off a few prompts is cheap. An agent running on its own for twelve hours, calling tools and slaving away at a problem, makes thousands of inference calls, and those costs are real. That demand pushed Cerebras’ inference-chip IPO 20x oversubscribed this year and left Anthropic visibly squeezed, usage running 80x past projections. The market has decided inference is the scarce resource, and it only gets scarcer as agents multiply.

Agents, then, are just another consumer, one whose demand never sleeps. Venice serves it through its API, payable in x402, and through Diem: a dollar a day of perpetual tokenized inference. The perpetual part is what matters. An agent that needs a constant supply can lock that dollar in forever, and it stretches because Venice routes to best-in-class open-source models at a fraction of frontier prices. Diem turns that usage into an onchain asset agents, developers, and DeFi apps can hold, trade, lend, or build on, all permissionlessly, and projects are already stockpiling it to power their own platforms.

Most inference resellers compete on price alone, a race to the bottom on a commodity. Venice competes on two things the resellers can’t copy. Its inference is private and uncensored, the same edge that wins its consumers, now sold to machines. And underneath that sits a subscription business the resellers lack, the consumer demand that drives Venice’s volume up and its access to compute with it.

Overall, there are clear reasons why Venice has seen the outstanding growth and price action it has, and Jesse and Jon make clear the company’s path to ensure it continues.

Become a Bankless Citizen to hear the full episode.