I think that his point is that glue is only as valuable as the product in which it's used.
In the case of finance, it's the fact that they operate as the middleman, getting money from one place to another. The product isn't just getting people together, it's providing a smooth way to move money around.
The other examples have a product where individual technologies were melded together to produce that specific product. They did much more than just take a few technologies, then bolt them together.
On the other hand, Summly seems like the technology should be the product. NLP is a very hot research topic, and when I first saw that the product went for $30 mil, I thought that they had some brand new NLP technique or algorithm that Yahoo was snapping up. In fact, several articles, like this NY Times article(http://www.nytimes.com/2013/03/26/business/media/nick-dalois... ) supported that view.
However, the truth is that they just took a pre-existing tech, and bolted it onto an iOS application. I can't comment about the app, since it's been pulled from the iOS store, but it does sound like they attempted to do something new with the interface, but that's about it. So Yahoo basically bought 2 engineers and a Siri NLP engine license, the later of which will undoubtedly become more expensive when the contract is renegotiated.