KorvaJoin the waitlist
Startups

Cheaper AI Means Little Unless the Automation Actually Works

Jev co-founder Diogo Almeida says only dependable background AI can turn lower prices into soaring demand.

AI image
The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch · 20VC: Jev Creator on Why Data Is More Important Than Compute | Will US Open-Source Models Catch Chinese | The Cost of Intelligence and Tokens: What No One Sees But Needs to Know with Diogo Almeida

Diogo Almeida, co-founder of Jev, a startup building AI models for software developers, was asked a blunt business question: if his tokens are so cheap, how does he ever build a huge business on them?

His answer was that lower prices alone do nothing. "The way Jevons Paradox tends to work, it's not like just magic, you make it cheaper and it gets better," Almeida argues.

Jevons paradox is the idea that when a resource gets cheaper, people find so many new uses for it that total spending rises.

In his version, a customer has a budget for AI. A cheaper model lets a stalled project finally pencil out, he says. But the project must then work reliably enough to create so much value that the customer tries intelligence on things it never touched before and expands the budget.

That reliability condition is everything, Almeida believes. He says that on the measure that matters most, true automation that runs on its own in the background, his own company and two other products he named are still at roughly zero percent.

The bigger prize, he argues, is not a coding assistant that helps engineers write software about 20 percent faster. It is software that does work quietly by itself, without a person checking each step.

If that kind of dependability arrives, Almeida predicts a "mega Jevons paradox." He expects decision making models to be used everywhere, with total use dwarfing today's text generating models, as AI shifts from something that throws unwanted output at people to something that improves daily life.

Getting there, he adds, will require more research and much more efficient service to handle the demand.

Questions people ask

What is an AI token?

A token is a small chunk of text, roughly part of a word, that a model reads or writes. AI companies charge by the token, so token price is effectively the price of using intelligence.

What is the difference between a decision model and a regular chatbot?

A chatbot that generates text mainly produces words for a person to read and check. A decision model is meant to choose actions, such as calling software tools or approving steps, reliably enough to run without constant human review.

Why is background automation harder than a coding assistant?

Background software must handle edge cases correctly over long runs, where one bad choice can corrupt records or trigger wrong actions. Current models can seem brilliant most of the time yet fail unpredictably, so companies keep a person in the loop instead of letting them act alone.

More in Startups