The system is still erratic in general

With both the generation of Thoughts and, most particularly, with videos, the system is still frustratingly erratic. I managed to publish one thought on interviewing Paul McCartney without issue. However, I’ve had three attempts at another on list-style thought without success, and in spite of the first version being publishable but then crashing when I tried to publish it (and then reporting the draft was lost).

The second and third goes were derailed by the AI not being able to follow instruction and/or producing incomplete thoughts. Image generation is also hit and miss.

With video, even though I’d uploaded a best-practice guide to my knowledge base, and provided a written brief tailored to it, I spent 3 hours trying, and failing to have a video produced, and again with several “it’s done, no it isn’t, wait just a minute more” prompts back.

Ultimately, at present too much time is taken up with these dead ends for the interactions to be enjoyable, or productive, and the system can’t as yet be relied upon to deliver.