Let's start with the bad. First, we couldn't get the Autobuilder build to work at all. The design is very finicky, as the grabbing claw isn't connected to a motor, and is completely reliant on the friction being just right to pick up pieces and release them in the right place. We simply couldn't get this to work. If the friction was enough to pick up a piece, it wouldn't release it. This was obviously very disappointing, and was only made more frustrating by a week of utterly worthless interactions with customer service via email, chat and phone, which ultimately seemed to conclude with a shrug and an "it's an inconsistent build." (For reference, the phone seems to be the way to go on the customer service, although the live chat was helpful once when a firmware update froze up late at night)
The second major problem is that the app is essentially crippleware, in the sense that it deliberately disables many of its programming features when you are trying to use them with one of the main builds. This is an artifact of the basic design, which for each build walks you through a series of activities which gradually introduce new features for each build. For each build, you can only access coding blocks after they have been introduced in an activity, but the result is that fundamental coding features like variables and conditional statements are completely unavailable in most builds. You can use them if you start a robot project from scratch, but then you can't access all of the blocks specific to the individual builds (of which there are many). This problem could easily be fixed by an update to the app, for instance automatically opening up all coding blocks on a given build once you have finished all the activities, but it is a huge design flaw, and makes the claim that the product teaches children about variables borderline false advertising.
Now for the (mostly) good news. In terms of play value, this has to be one of the best, if not the best Lego set for the money. The fact that it comes with five builds (four if we ignore the Autobuilder) and an extensive set of activities for each one is a great start, and of course it allows building and programming a robot from scratch as well. But it also includes several builds that can serve as the underlying base of a build-your-own robot, including a walker and a wheeled vehicle. We haven't tried these yet, but they look like they will be great gateways to completely from-scratch designs. I think the cat was my favorite build, partly because nearly all its various activities integrated well and could naturally be combined into a single complex program which for instance randomly chose between several options for what the cat would do next. The guitar was a pleasant surprise, mainly because it behaved so differently from the others -- it demonstrated that the motors could be used as physical input devices rather than to move parts of the robot. Vernie and MTR were cool but a bit underwhelming, and in terms of activities both felt a bit like a bunch of random conflicting ideas tossed together. I did end up with a sense overall that the app team ended up underemphasizing actual physical interaction with the robot, so that my kids spent a lot more time dealing with the iPad than looking at or interacting with the robot. But that's not to say there weren't a solid number of opportunities for interaction with each model.
Minor flaws included a buggy app with a tendency to crash, and sensors that weren't always reliable (mistaking similar colors, or behaving slightly inconsistently when measuring distance). It seems like more careful testing of the builds and activities could have mitigated this even without hardware changes. The app also has an annoyingly persistent insistence on avoiding words, which makes essentially no sense in the context of a product recommended for ages 7-12, and results in nearly non-existent documentation that makes a lot of its behavior far more mysterious than necessary.
It goes through batteries fairly rapidly, so a good set of rechargeables is recommended, but in all the builds it's easy to swap the batteries in and out, so this was no problem for us.
Overall, it's frustrating in that it could easily have been a total home run, but the two major issues described above create a lot of unnecessary angst. Nonetheless, it's already provided far better value for money than any of the other Lego products I've gotten for my kids, and I expect it will continue to be a major activity for the foreseeable future.
