Matthaus is an investor in Rasa NLU and one of the sharpest persons I know in this area. It is always a pleasure to get a peek in his mind, like here: How four of my bets on Rasa have evolved – or why the enterprise AI market is still underserved
I get a lot of my ideas from looking into what devs I respect are building. Looking into that it was pretty clear that what was being built was very, very simple.
Matthaus is right here that the initial efforts towards conversational technology were really simple. That has also always been my recommendation since the marriage of technical challenges and design direction is difficult enough to pull off, let alone really push the envelope in any one direction.
I don’t think that problem of collaboration has been solved or ever will, but just that tools like Rasa take a probabilistic approach to the problem, which either changes what it means to design a conversational system or wholly sidesteps the problem (whichever way you look at it).
The market is way bigger than I thought.
This is something I’ve noticed as well from a different angle. I’ve been waiting for the initial batch of services to die out. I had half assumed most of them had already done so when I resumed giving talks a year ago. Turns out that they really didn’t. Despite the disillusionment, this is a thing that is still happening everywhere. Maybe not exactly the way some people thought or wished it would, but still.
The combination of fast product/hard science teams is very rare.
I don’t have an answer to this though I would say that this kind of team chemistry is rare in general. The special demands in this space on team members, of course, make it even more difficult.