#53 - Be Wary of Solving a Small, Rare Problem - Des Traynor of Intercom

**Craig Cannon** (0:00)
Hey, what's up? This is Craig Cannon, and you're listening to Y Combinator's podcast. Today's episode is with Des Traynor. Des is the co-founder of Intercom, and he's also given a bunch of talks about product.
So many of you guys requested him as a guest, so Des came on to talk about product, and he also talked a bunch about how he's changed roles since he co-founded Intercom, and he shared a bunch of lessons that he's picked up along the way.
All right, here we go. How did you meet your co-founder and decide to get going?

**Des Traynor** (0:29)
Sure. So I was originally a computer science student, and I started a PhD. My PhD was an attempt to see if we could automatically measure how good a programmer is, basically. So I put them through a variety of sort of simulated code and asked them to say what the output would be.
And we had some really interesting findings from that research, but I got bored. I got bored for two reasons. One is that some of the findings were slightly uncomfortable for some universities because I was basically saying that, hey, this person you're saying is getting a 65% score. I actually don't think they can program because they can't tell me what this code outputs.
People don't want to hear that, obviously.

**Craig Cannon** (1:03)
Just because they figured out how to take exams well?

**Des Traynor** (1:05)
Yes, precisely. So typically, the way, and this is true worldwide, I think the way people in universities would mark a piece of code is they say, all right, we'll give them three points if they have a for loop and two points if they have an if statement. And people just learned that off. So the phenomena is that template-based marking encourages template-based learning. So people know what a solution should look like, but they have no idea what it does. It's like learning off Latin or something. So I was kind of that finding became more and more comfortable for universities. And as I said, hey, I'd like to do some years study on your students.
Here's my hypothesis. You're like, no, thanks.
So I started writing a blog instead because that was the alternative to publishing, was just to write about this stuff. And then I started writing about design and usability. One day somebody commented on my blog. His name was Owen McCabe, and he was organizing this coffee morning in Dublin city for all the people who worked on the internet, which is a small group. And this is like 2006
And I went along and actually a funny story is I met Owen for the first time there, who I went out to start Intercom with. I also met my wife for the first time at that same meeting.

**Craig Cannon** (2:10)
Wow, I fell for you.

**Des Traynor** (2:11)
Yeah, it was exactly a good day's work.
So I actually went and took a different job with a consultancy.
And then for one year, me and Owen kind of kept in touch. And then one day, he mailed me and said, Hey, I'm starting my own thing. Do you want to get involved? So we started working. We had an agency called Contrast. It was like a design agency. And then we were following the 37 signals model. We wanted to have a side project that we'd eventually pivot to.
So our side project was a tool called Exceptional. It was a Ruby on Rails based error handler.
And it was like alert developers and notified them, hey, this piece went wrong, and here's who was affected.
And so that was the normal trajectory. And then one day, we were getting really frustrated trying to communicate with our users, because we had thousands and thousands of people using the product. But the product was frequently performed slowly or fall over. And the state of the internet back in 2010 wasn't good. So the infrastructure wasn't there. There was no stripe. If we wanted to mail all of our paying customers to apologize for some downtime, it was like, log in to the PayPal dashboard, get an export of all your subscriptions, run that through some pearl to extract the email addresses, import the email addresses into Campaign Monitor Mailchimp, send out a notification, get all the replies and all the auto replies and all that shit back into your own email.
Resurrect your email after two hours of going through it, trying to fix it up. And all of that, we had to do every single time we wanted to say anything to our customers. That was the norm. So one day we had this little message pop up in the bottom right-hand corner of the product saying, hey, we're sorry we were down yesterday. Here's the changes we made. Read our blog post.