Perplexity Did What!?
In early November, I was hunting for internships when I decided to see just how good Perplexity had gotten. I gave it a well crafted prompt, with the expectation it would give me disappointment and a few links - I got a spreadsheet. The first hint something was off was a small, spinning “loading wheel”, next to a tag saying “Reasoning: Beta”. I had never seen this before.
The prompt itself wasn’t random. When I wrote it I was sitting with John in the garden behind the library. He’d scored what I can only assume was a natural 20 on persuasion, because a request for sample data had landed him a dataset of 700 Bay Area companies hiring interns. I had filtered it down to 13 AI startups, because, well, those are the jobs I am looking for. In my opinion, that sounds like a great stress test for Perplexity.
What happened next wasn’t normal. Instead of a list of 3 links, and an incomplete answer going over the first 2-3 companies, Perplexity went into overdrive. The spinning wheel felt like it was taking forever. When it finally stopped, there was a detailed spreadsheet, with 65 citations, giving me the exact structure I had asked for.
Actually, I Manifested This
In my mind, luck isn’t something that happens to you. Luck is just a label that’s put onto you after something happens to you they deem “lucky”. The 45 minutes spent on making a rubric wasn’t a plan- it was a question. In making the rubric, I was asking “what if this works? What if Perplexity gives me what I want?”
Personally, I don’t want to work at a company I don’t believe in. I want to know about their business model, who they are selling to, their funding– I want to work at a place that isn’t scamming people, and I want to work at a place that I think might succeed. Of course, I would also want to know about tech stack, and what roles they have open, but that’s easy to figure out from just reading the job listing.
I didn’t choose to spend those 45 minutes because I had any clear expectations. I just wanted to know how powerful perplexity had gotten, and I wanted to sort my thoughts. I wasn’t preparing, I was playing. And by playing, I found an opportunity I could have never planned for.
Hackathon: START!
It’s D Day, and here I am, sitting on the floor in the corner of the main room (yes, there was a perfectly good desk available - no, I will not be taking questions about this decision). Gev finds me first. As most of the City College delegation gather, I rope in two new members- Sean, and Ilia.
I have a team. Quickly, before my opponents realize the game, we commandeer the massive whiteboard. I have never touched reflex before. I’m not worried– it’s not my first hackathon, and I have gotten pretty good at learning as I go. Though a bit vulgar, my motto has always been “fuck around and find out”.
The plan is simple- I take the front page, making it basic to save energy for fighting integration fires. Gev gets ownership of the company profiles, the most important component. She has worked with Reflex before, so it’s a great fit. She delegates the jobs tab to Ilia, and is there to mentor him. Sean gets the database, and although I know he will need help, at least he has set up a database before.
Hindsight is 2020
In the blink of an eye, the hackathon was over. The integration fires came, and we beat them. The presentation came, and we gave it. We walked away with a trophy for “most useful/impactful idea”. All of this because Perplexity just so happened to give me a spreadsheet with 65 citations.
Looking back, I can only guess it was A/B testing - I mean, why else would I have access to a feature for exactly one day and never see it again? That “Beta” tag was there for a reason. When I was writing the prompt, I wasn’t plotting. Taking advantage of this A/B test wasn’t some grand scheme. I was just playing around with tech I thought was cool, asking questions whose answers I genuinely wanted.
Or maybe, like Marta Cabrera from Knives Out, I was creating beautiful shapes. In creating the prompt, I was asking two questions: “Can Perplexity do this?”, and “What if Perplexity can do this?”. Somehow, everything just.. aligned. The project ended up being perfect for mentoring newer programmers (which was actually one of my goals for SCC Hacks). I didn’t plan that. I was just being proactively myself, and the pieces fell into place.
This is how I see opportunity: don’t try so hard for specific things, just have fun and keep asking interesting questions. All I did was ask the question, but If I had never asked the question, I would have never gotten the answer. Eventually, you will ask a question that has an answer you like, and the world will call you lucky.