GoInstant

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

~ the goodbye ~

Halifax is not where this normally happens. Jevon MacDonald and Gavin Uhma built GoInstant in Nova Scotia and took it to the TechCrunch Disrupt stage in San Francisco on 13 September 2011, where the demo was the kind that needs no explanation: two browsers in two different cities showing one page, and when one person scrolled or typed the other one saw it immediately. No plugin and no install. A line of JavaScript on the page you already had.

The money arrived the next day. $1.7 million from a set of Silicon Valley funds that do not usually look at the Canadian Maritimes, announced by the provincial government as the win it was. Then the product widened. What started as shared browsing turned into a general backend for real-time applications: an API that kept a chunk of state synchronized across every connected client, so a developer could build a multi-user thing without operating the socket layer that makes multi-user things hard.

Salesforce agreed to buy the company in July 2012, reported at around $70 million, less than ten months after the public launch. The CBC headline called them Halifax whiz kids, which they mostly were. The co-browsing technology went where you would expect, into Salesforce's customer service products, so a support agent could see the form a customer was stuck on. GoInstant the developer platform kept running alongside it for two more years, with its own docs and pricing, and a WebRTC integration that Salesforce's own developer blog was still writing tutorials about in February 2014.

~ the chronicle ~

~ the halifax exit ~

For Atlantic Canada this was the number that reset everyone's expectations. The region had a startup scene measured in single-digit millions, and a four-person team at a Halifax address sold for a reported seventy within three years of starting. Both founders stayed in the ecosystem afterward and a good deal of what happened in Halifax tech over the following decade traces back to that sale. The company's actual product gets remembered locally as a detail of the exit rather than as software, which is a little unfair to software that worked.

~ the tombstone ~

Whatever was posted internally, the public version was eight lines of HTML. By October 2014 goinstant.com had been replaced by a page with the logo, a heading reading "The GoInstant Service Has Been Discontinued", one sentence saying "Effective August 31, 2014, The GoInstant service has been discontinued", and a link to a blog post titled "time to say farewell" that the Internet Archive never caught. Under it, a copyright line for 2014.

The archive has a small joke in it. The full marketing site, with the pricing and the sign-up button still on it, was being crawled on 9 September 2014, nine days after the service it was selling had stopped existing. Nobody had got around to the homepage yet.

~ what developers lost ~

This one lands at a specific kind of distance for me. By 2014 I had been writing software for years, but on the embedded and vision side, so a JavaScript sync layer was a thing I read about rather than a thing I shipped. The people it mattered to were the web developers who had wired an SDK from a Canadian startup into a product and then got two years of notice that the vendor's owner had other plans for the team.

It is the standard risk of building on a young company's API and the standard outcome of a successful exit. Salesforce bought engineers who understood real-time collaboration in a browser, which in 2012 was a small and expensive group of people, and it put them on Service Cloud. Everything else attached to those engineers was overhead on an acquisition that had already paid for itself.

@}-->--

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

« previous  ·  random  ·  next »