We can genetically engineer bacteria to mimic the missing pancreatic cells, and it's not too different to the way most insulin is produced as all that's new is the system to stop producing insulin when blood sugars are already low enough. However, if you put them in a person, the immune system attacks the bacteria, so they need isolating. To do that, we need a membrane that lets sugar in and insulin out, but doesn't let antigens or live bacteria out, and doesn't let immune cells in. Even if the bacteria are held in place, if immune cells can get in, it's no better than a pancreatic transplant as you'll still need immunosuppressants, and they're generally worse than dealing with type one manually. Development of the membrane keeps hitting unexpected hurdles, so artifical pancreases are still unable to start trials, and then they might take a decade.

There are other approaches, e.g. using electronics to control photosensitive insulin producing bacteria, but they don't have any advantages (the membrane still has to let sugar in so the bacteria can eat) and have more things that can go wrong.