Cue

» see greplin.com as it lived, on the wayback machine

~ the obituary ~

Cue was built for a life made of flight confirmations and restaurant bookings, and I was not living that life. That is the honest distance on this one, and it is the same reason the product worked so well in demos filmed in San Francisco.

It started in 2010 as Greplin. Daniel Gross and Robby Walker founded it with Shai Magzimof out of Y Combinator, Gross barely out of his teens, and the observation behind it was correct: by 2010 your own material was scattered across half a dozen services that all had search and none of which searched each other. Greplin indexed your accounts and gave you one box over the whole pile. Sequoia put four million dollars in during 2011. Paid tiers ran five dollars a month for the first chunk of indexed storage and fifteen for more.

In June 2012 it became Cue, and the box became an answer. Instead of waiting for you to search, the iPhone app read the same accounts and assembled the day ahead. Your flight was in there because the airline had emailed you. Your dinner was in there because the reservation had. It was aimed squarely at Google Now, which had shipped a few weeks earlier and was doing the same trick from inside Android.

Apple bought the company in October 2013 for a figure reported between thirty five and fifty million dollars. The shutdown notice went up on 1 October, before the acquisition was even confirmed publicly on the third. Premium subscribers got refunds. The app stopped working, the site stopped answering, and there was no wind-down period at all.

~ the case file ~

~ what it was reading ~

Gmail was the important one. Everything else fed it context, but the mail is where the structured facts live, because companies send confirmations in machine-shaped templates and have done for twenty years. A booking email is a small database record wearing a paragraph.

Which is exactly why the product had a geography. It worked if your inbox filled up with the right templates, which in 2013 meant American airlines and American restaurant systems and American parcel carriers. In Ankara the mail that arrived was not in that shape. Turkish Airlines and a bank and a phone operator, all of them sending things the parser had never been taught. The assistant category was regional long before anybody said so out loud.

~ the pivot that got it bought ~

Search over your own data is a feature, and features do not raise second rounds. The move from Greplin to Cue converted the same index into a product with an opinion, which is the version somebody will buy, because a company that can turn a mailbox into a schedule has solved the part of the assistant problem that is not voice.

Apple had Siri, which could hear you and could not anticipate you. Google Now could anticipate and was tied to Android. Buying Cue was the shortest path to the missing half, and Apple bought the engineering rather than the userbase, which is the standard explanation for a shutdown notice that goes up before the deal is announced.

~ where it ended up ~

Robby Walker stayed at Apple for around nine years and ended up running Siri and the information intelligence work underneath it. Daniel Gross went into Apple's machine learning effort and later moved into investing and his own AI companies. The team was the asset and the team went where it was bought to go.

The product logic survived too, in the proactive suggestions that started showing up in iOS a year or two later. Your boarding pass surfacing on the lock screen at the right hour is Cue's argument, shipped under nobody's brand. That is the usual afterlife for an acquired assistant. The idea gets absorbed and the name gets deleted on a Tuesday.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »