Ambitious Boston-based startup Nara is developing artificial-intelligence technologies such as neural networks designed to provide consumers with information on topics of all sorts tailored specifically to their interests. But its first service, which it’s launching today in beta form, just does one thing. You tell it about restaurants you like–it covers seven cities at the moment–and it recommends new restaurants for you to try.
I tried the service a couple of times, inputting different sets of favorite restaurants in the San Francisco Bay Area each time.
In test #1, I told Nara I liked a French-Vietnamese restaurant, a Chinese restaurant, and a tiki-themed pan-Asian one. (Oh, what the heck: The tiki-themed pan-Asian one was the famous Trader Vic’s, but Nara didn’t list a neighborhood, so I wasn’t sure whether I was choosing the flagship location in Emeryville or the outpost in Palo Alto.)
Nara returned a never-ending scrolling list of restaurants which I could refine by marking additional ones as favorites. How’d it do at identifying ones I’d like? Without going out and trying them all, it’s tough to know. I did notice that one of its first recommendations was another Chinese restaurant which I do like and hadn’t mentioned. But it also listed establishments that seemingly had nothing to do with the ones I’d picked except for geographical proximity, such as a place called Extreme Pizza. And in many cases it didn’t tell me where a restaurant was until I clicked on it. (It knows about ones all over the Bay Area, but only seems to have neighborhood information for San Francisco proper.)
In test #2, I chose three restaurant/bars in San Francisco’s SOMA neighborhood. Nara responded with other eateries in SOMA–lots and lots of them, in all cuisines and all price ranges, from fancy joints to a taco truck.
Nara has some nice features: You can save promising venues to a grid and jot notes for later reference. It also hooks into OpenTable so you can make reservations; that’s part of its plan to make money. But when I tried it, it didn’t seem to provide more useful advice than I’d get if I simply perused Yelp and looked at what other folks like. Given that the service is so new and still in beta, that’s understandable, but it seems to raise its own bar by immediately declaring that it’s found the restaurants “that suit your palate best” and urging you to book a table. Lowering expectations might be a better strategy.
The site plans to add hotels, entertainment and shopping recommendations soon. It’s worth keeping an eye on.