Packing chemical vials probably won’t get a standing ovation at a technology conference.
We’d still rather hear about the job.
Verne Robotics, founded by Neil Nie and Aditya Jha, is developing robot arms that learn manipulation tasks from demonstrations. Its public launch describes Nemo3, a two-arm system, and a deployment packing vials at ABClonal.
The company says it went from unpacking to automating that task in four days. That is its account, and the setup time is exactly the part we would want to examine with a customer.
Because getting the robot to do something once is a very different expense from getting it to do something usefully every day.
There is the demonstration data. The workspace. The objects. The handoff when a person needs to intervene. The changes a customer makes after you thought you’d finished.
Verne also describes an hourly payment model. That creates an interesting commercial question: can the team get a robot working quickly enough, and keep it productive long enough, to make the economics work for both sides?
That’s a much more useful question than how many tasks a model can theoretically learn.
Our read is that a narrow packaging job is a sensible place to test the larger ambition. A customer has an existing process and a reason to compare results.
We want to see what happens on the next deployment. Does the learning carry over? Does support become easier? How often does the workflow change?
Tiny vials. Repeated movements. A very specific piece of work.
There are worse places to start finding out whether flexible robotics can become a business.
Budera Worth Watching. The companies we think you’ll wish you found earlier.
