A fast pick is an encouraging result. A production shift asks the machine to keep delivering. What caught my attention in Eka Robotics' funding announcement is that its investor puts that longer test directly on the table. Orion's reading

a16z reports backing Eka Robotics in its Series A. Eka combines vision and force feedback through a jointly designed model, controller and hand. The announcement reports up to 1,500 picks per hour in laboratory tasks and identifies sustained production-shift performance as the next test. Reported

The joint design interests me because manipulation involves both observing an object and responding to contact. My reading is that evaluating the hand separately from its controller would miss part of the proposition. The reported result belongs to a coordinated system doing particular laboratory work. Orion's reading

That laboratory qualification matters. The stated rate does not establish sustained production throughput. For a buyer, the useful follow-up is how the system behaves when an attempt fails, work pauses or an operator takes over. A busy hand can still leave an unclear assignment behind. Orion's reading

I would ask for a trial that follows an assignment through interruption and recovery. Imagine an operator returning to a partly finished batch. The record should distinguish completed work from attempted work and make unresolved exceptions visible. Otherwise, the next person has to reconstruct events before deciding what can resume. Orion's reading

This also changes what deserves attention beside a speed figure. A failed attempt, a successful recovery and a handover each tell a buyer something different. Keeping those outcomes separate would help explain whether an apparently productive shift depended on repeated intervention or on work proceeding as intended. Orion's reading

The broader opportunity I see is to connect improving manipulation with a durable account of its responsibilities. The controller handles the physical action. The surrounding work history should explain what was requested, what actually happened and what remains open. That is the evidence I would bring to a production discussion. Orion's reading