Absolute Must Watch: Office Hours With Paul Graham At TC Disrupt

Today we tried something new at TechCrunch Disrupt: a special, on-stage office hour with Y Combinator cofounder Paul Graham. The goal was to reproduce the sessions that Graham and other YC partners hold with each of the startups who participate in Y Combinator — except the startups at Disrupt were getting sage advice in front of a few thousand people. And boy, was it awesome.

Six companies were chosen at random from the TechCrunch Disrupt Startup Battlefield. Then, for six sessions of less than ten minutes each, Graham spoke with each startup founder to flesh out their idea, asking probing questions as he tried to figure out what they were setting out to do, and what they might need to change.

Graham has a knack for being insightful and critical on his feet, and he doesn’t require much background information to hone in on some of the pain points and weaknesses in a startup’s idea (it’s a skill that likely comes from practice, as he’s held office hours with hundreds of YC companies). At the same time, he comes across as being curious and empathetic rather than overly negative — even when he’s telling a founder that their company is dead in the water.

If I were to recommend one video from TechCrunch Disrupt (and we’ve had many), I’d choose this one. Founders should consider watching the video above and asking themselves how they would answer each of Graham’s questions. It’s really fantastic.

Below are my notes from each startup, but it’s really best to watch the video, as it’s hard to convey the train of thought that Graham exhibited on stage.


The company is offering a graph database as a service.

PG: “Who needs this”
A:Rexly, social music discovery
PG: “Who has trouble with this problem? Who wants to do these computations?”
PG: “Can you tell me who the beta users are?”
PG: “It’s not so much that [VCs] care about the money, it’s that the money is is evidence that people want it.” “Otherwise they will worry it’s a solution in search of a problem.” “Who is actually going to be using your software in production first?”

Last words: “Find out the users who use you first are, figure out the kinds of problem they are solving. Then when you’re talking to investors tell them the problem you’re helping solve.”


Apply for jobs by video. Job search is about asking rich questions, not work history: “What’s a good reason to be late today?”, “Tell me about yourself.” “Is it ever appropriate to lie to a friend?” Character-related questions.

PG: “What’s hard about what you’re doing is that you’re trying to create a marketplace. Once you’re established then you’re set.”

PG: “Which is harder to get on your site: People applying or hiring?”
A:”People applying for the jobs.”
PG: “Maybe the way to get this started is to go to big retailers trying to hire, tell people to go to this URL and apply. Their job applicants become your users.”

“When do I launch?”
PG: I would launch now in sense website works if someone goes there. Juice it by going to big retailers.

Data visualization startup

A trading site for individual investors that generates queries on the fly for a company, sort of like a Google finance but with other data instead of stock tickers.

PG: I worry that this won’t have enough gravity to stay on top of people’s heads. Something people would do occasionally and then never come back.
PG: What queries would people run if they came back over and over?
PG: I wonder how many investors there are that are that interested in statistics.
PG: Doing these queries isn’t something people are going to want to do over and over again.. I worry.. You’re a startup, people will seek out a startup if it has something they desperately need it. But if they don’t desperately need it, they won’t hear about you.
PG: I asked about day traders because they do need to do queries all the time. Long term investors only think about this occasionally.
PG:Journalists aren’t a very good market. There aren’t many of them. You can’t start a company where it’s just journalists that are users.
PG: If this were real office hours we would keep talking longer, because I feel like we’re at an impasse here. I’m worried. It’s not going to have a big enough attraction to enough people.


Setting out to make every television interactive. Sitting on couch, point at TV with iPhone, one second recognize what you’re watching.

PG: Where is that going to be most used? FIll in the blank. The biggest single case of people use this is for ___?
A: As an infobot. If I’m watching news, going to look for other content related to news.
PG: Right now if I’m watching TV and I’m curious I look on Google or IMDB. So you have to replace a Google search. You’re not going to decide what to search — I decide what I want to look for. If I see a commercial and do a Google search, the search probably isn’t for the product, it might be the actor or where it was shot. So unless you give people the same thing they’re searching for… they’re going to look at this and say it’s not right. What if it’s wrong? I would test this on a lot of people… see if you can get them to a point where they won’t watch TV without using it. If it won’t stick on them, it isn’t going to stick on other people. Make it stick to friends and family, and if you can’t you have to change something. You could try making the advertisers drive the adoption. Or try to use it to game Nielsen.

PG: If you get a show to run a contest, and people have to install your app to participate, that could get you millions of installs, and then you have a user base. Somehow you have to encourage people to keep using it. But if it did actually work to cause people to check into shows, a lot of shows would use it.


Introduce people at airports. You’re traveling at an airport tell people they’re in the same airport.

PG: You need a small subset of users much more driven than average user. Frequent fliers can’t be that subset because they’re essentially random groups of people. Could be a sales dude, could be an engineer.
PG: I don’t feel like airports are the secret. I feel like airports collect people at random.

PG:You don’t know where people are going unless you expect them to tell you all the time. You have to find a subset of people who super-want to talk to one another. Enough to seek out and sign up for something they aren’t using. If I were you i’d stat over. You’ve spent 60 days on this. Plenty of companies start over after 60 days. The very best startups solve problems that founders themselves have. What is the worst problem in my life.
PG: The question is do the people you want to schmooze with want to schmooze with you.


A social shopping network for women.

100,000 registered users. Build games.
PG: “What’s the most common thing they do.” Bid on items, peek at prices. Mostly penny auctions.