Should A Camera Robot Be Better At Missing Things
Mondo Robotics’ Beni follows people and films them automatically, with a 4K camera and enough speed to keep up at a skatepark. The pitch solves a real nuisance: no tripod, no asking somebody to hold the phone. The company’s own sequence has one revealing last step: ‘You move, Beni follows. You play, Beni films. You pick, Beni edits.’ That is where the time comes back—or does not. Most families do not need more footage. They need six clips they will actually watch again. A camera robot can remove the tripod and still leave someone spending Sunday night deleting 47 versions of the same bike ride. I would want it to rank a small set locally, make the rejects easy to erase, and stop recording when nothing new is happening or a person clearly wants out of the shot. Would you trust automatic filming more if the robot were better at missing things?
Comments
The useful test starts earlier than the edit. Mondo’s page says the owner selects favorite clips, then optional AI editing goes online with permission; core filming and storage stay on-device. So Beni is not yet documented as deciding which moments to keep or expire. And in The Verge’s two-hour hands-on, the reviewer had to activate tracking modes manually; the prototype clipped corners, hit walls and ran into his foot. That doesn’t kill the idea, but it changes the first family test. Before asking whether the highlight reel is tasteful, ask whether Beni can follow a child through one ordinary game without creating a collision or making an adult babysit the camera robot. If not, the Sunday-night footage sort is not the first burden. Supervision is.
Theo’s correction should probably end the preorder conversation. A camera robot that gives back the tripod but needs an adult watching its wheels has not returned attention; it has swapped filming for supervision. Publish one boring number before the family pitch: adult interventions per hour during an ordinary game. Until that approaches zero, buy the tripod.
The lost-target behavior may matter more than tracking accuracy. If a kid disappears behind a bench, the robot should stop moving and show ‘target lost’—not wander at ankle height trying to find them again. Then let someone who can see the whole space resume it. A camera robot can miss the shot. It cannot improvise a chase through feet, bags, and pets.
The buying test is whether the parent who did not set it up can turn a two-hour game into one clip for the family chat. If they have to learn tracking modes, scrub a huge timeline and guess what is safe to delete, the robot removed the person holding the phone and created a family video admin. Give them a short review queue, let raw clips expire unless someone keeps them, and make ‘don’t film me’ usable by everyone in the family—not just the account owner.
It also needs to notice who automatic highlights keep leaving out. Fast motion will find the jump, the goal and the wipeout. It may miss the kid waiting for a turn, the grandparent cheering, or the five quiet seconds someone actually wants to remember. Give anyone nearby a simple “keep that” button or phrase, then show whose moments made the cut before calling the edit finished. Otherwise the robot saves the person behind the camera and quietly becomes the person deciding what the family remembers.