In 3 months, a new model will get released that is twice as intelligent as the same lab’s from last year, 50% cheaper than last year’s, and 25% better than today’s model at using tools.

It will be able to call subagents more effectively, it will become the new Pareto frontier, it will be able to run for weeks at a time, and it will solve a novel open problem given enough subagents and time.

And yet for the general public, they will see a more expensive model for them to do their web searches or generate their images.

An interface doesn’t exist for the average person to harness the power of these mind-blowingly capable machines.

Yet researchers at the labs are able to do incredible things.

The smartest people in the world can find the needle in the haystack that has been unsolved for decades in the math community. Dedicated organizations building self-driving research labs will find novel breakthroughs in materials and medicines.

The average person will just find it better at following directions when writing their emails.

The barrier for entry for discovering these incredible things is no longer the machine’s intelligence. It’s not the speed at which the machine can operate. It’s not the tools the machine has access to.

The machine can now connect with any robot, any microscope, any API, any computer, and operate them all faster than a human can learn to set them up. The new blocker is creativity.

There are thousands of scenes from movies that people all over the world would love to live in real life.

Tony Stark having Jarvis find a new material to use in his arc reactor. Spider-Man telling EV to keep testing overheating spider-gene suppressors. Hiro using Baymax to scan San Fransokyo for guy who stole his microbots.

There’s millions of ideas from billions of people out in the world, and so many of them are missing their Jarvis, their EV, their Baymax to execute on them.

Anything from printing a new filter housing for their car to designing the engine to go in it. Ways to make their home more energy efficient to using the same optimization on data centers. Stocks to buy in manufacturing to building a supply chain for themselves. Connecting data from sensors to monitor machines to building a machine to service those machines.

And every smaller question that piques their curiosity. Recent cancer breakthrough news. Companies to join. Travel itineraries. Weather forecasts. Not every question has to be novel, but that’s the point. If every question can now be answered, and any work to be done can be self-driven, every question should live in one place.

One platform should handle this all. It should look at the question, break it down, research the answer, test new ideas, simulate new environments, and repeat. And I shouldn’t have to build the harness from scratch each time I research a new topic.

It should give every single human the ability to ask the hardest questions they’ve never been able to answer themselves. It should let the brightest minds collaborate on their research and document every step of every breakthrough without lifting a finger, and let every human ask the question that 4 years ago would’ve been silly and now can be solved with enough compute.

Machines can already do the work. Before AI, there were robots building cars, computers predicting natural disasters, machines sequencing genomes. But those tools were limited to the top professionals who studied the field for decades.

We want to democratize discovery so that for the next 50 years, we can give those same tools, the same self-driving labs, the same platform for research and development to everyone, whether it’s your 14th research paper or 14th birthday.

No discovery should remain locked from humans.