Delivery Reliability Favors Ground Robots

This is the core economic test in last mile delivery, because a system that looks cheap on average can still be unprofitable if the rare late, failed, or mishandled orders trigger refunds and break customer trust. Coco is built around tightening that tail. Hot food gives only a 15 to 20 minute service window, and once failure rates drift from nearly perfect to merely good, the profit from many successful orders gets wiped out by a small number of bad ones.

• Coco lays out the math directly. If contribution profit is about $1 per order, then 100 orders produce $100, but two failed or badly delayed $50 orders can erase that entire profit pool. That is why moving from 98% to 99% and beyond matters more than small average cost improvements.
  1 sacra
• The practical issue is not just completion, it is edge cases. A robot that stalls and forces one operator to intervene can delay many other active trips. Coco therefore targets at least three nines over time, using autonomy plus one to many teleoperation so operators can watch multiple robots without creating new delays.
  1 sacra
• This is where Coco draws the line against drones in cities. Coco argues ground robots can keep running through rain, snow, and wind on existing streets and sidewalks, while drone operators still face harder constraints around landing space, hover time, charging, and aviation approvals. Drone operators like Wing and Manna show strong reliability in their chosen suburban conditions, but they also organize service around narrow operating envelopes and hub economics.
  1 sacra 2 sacra 3 sacra

The next phase of autonomous delivery will be won by the systems that make rare failures boring. Ground robots are pushing toward that in dense urban food delivery, while drones are pushing toward it in suburban, hub based networks. As both improve, the market is likely to split by where each mode can hold near perfect service without blowing up cost.