I'd argue that the weeds were a type of technology used in that role: they were using something that sent down probes in to the ground slightly shallower than t

I'd argue that the weeds were a type of technology used in that role: they were using something that sent down probes in to the ground slightly shallower than the grape roots, and had a visual change when the levels dropped below a certain amount. In fact, the weeds have some benefits: they average over a small patch, have good volume coverage, likely have a water loss correlated with the factors that cause grape water loss, etc. If you view plants as evolved machines, then it makes sense that weed with certain properties (such as not strangling the vines (too much)) would make really good sensors. Trying to replicate that, as you point out, is a really hard challenge, and only matters if you're doing something complicated with the data (such as integrating many sensors in to an automated system that locally adjusts the levels).