Swipe Left on Swiping: How I Built a Matchmaking Website Called Intention
Another episode in my ongoing series "Arguments with Algorithms" - where law meets experimentation with AI - this time with coding tools.
My Crazy Idea
I'm sitting in a bar in Mendocino, having what I thought would be a casual conversation with two women from San Francisco about our dogs. But instead of typical bar small talk, they're telling me about their dating horror stories – the endless swiping, the superficial matches, the complete waste of time that modern dating has become.
That's when – like a mad scientist in a bathtub yelling eureka – I looked up from my pinot noir and said, "Hold up. This is insane. What if instead of mindless swiping, I was a matchmaker that took the time to understand your values, and then gave you like 5 dates? And used AI to match you. And people got dinged for ghosting?"
They looked at me and said, "Hell yeah! Sign me up." No joke, one of them said she would be my biz dev head if I actually built this startup. But that could’ve been the wine talking.
That conversation in Mendocino became the genesis of Intention – not born in some Silicon Valley garage, but in a real moment of human connection - more like frustration, fueled by good wine and the audacity to think I could solve a problem that to my mind requires more than an algorithm that sends you cute guys to swipe at (some weirdos as well, I’m told).
I think dating and building a relationship is deeply personal requiring what lawyers may call a ‘human in the loop’. A serial startup builder I was having coffee with said to me smirking - “This is your Indian aunty talking” - I actually think he’s right. Maybe it is, I’m certainly the right age for it (shrugs). Imagine if Seema Aunty was less (well herself) and had AI to help her do matching (beyond - you are from Guyana - he is also from Guayana). Btw if this reference is throwing you off, you really need to run and watch Indian Matchmaking on Netflix.
I also think there are various aspects to relationship building - and various papers and psychologists that have already done the work here, so why don’t we ask questions to get to know those aspects based on say, the Big 5 Personality Traits or Attachment Theory? And AI could learn that make an algo right? and do the matching - maybe?
Which is a long winded way of saying that I decided to create an app with a premise that was pretty simple: create a matchmaking platform that prioritizes genuine compatibility over surface-level attraction. Instead of the swipe-fest that dominates the dating app landscape, I wanted to build something that could actually help a matchmaker/human find their client their person.
High level - the process needed to be designed to get people meeting in person, not endlessly texting. So a human does intake - asks a bunch of questions to feed into AI tools that helps match for compatibility. 5 dates a month. Users link their calendars and actually meet face-to-face – no more of this "let's chat for three weeks and then ghost each other" nonsense that plagues dating apps.
After each date, users share voice reflections through prompts that help the AI learn what actually works for them. It's like having a matchmaker who gets smarter with every interaction, constantly refining future matches based on real-world feedback. Then at the end of 5 dates, there's a personalized insights report plus ongoing matchmaker support to help people continuously improve their dating approach.
Anyway, this gave me the perfect excuse to try something the cool kinds were calling ‘vibe coding’.
Vibe Coding 101
Vibe coding isn't just about using AI to write code - it's about having a conversation with AI about what you're trying to build. Instead of thinking 'I need a form that collects data,' I'd tell Claude 'I need something that makes people feel comfortable sharing vulnerable information about their dating preferences.' Then we'd work together to figure out the technical approach that creates that feeling.
Here's the thing about being a product lawyer who codes: you approach problems like you're building a case or project managing a launch. Every feature needs to serve a purpose, every user interaction needs to be intentional (pun intended), and everything needs to work exactly as promised.
Biggest hack obviously, is that I used Claude to generate most of the code. That's right – I had an AI teaching me how to build an AI-powered matchmaking platform. The irony is not lost on me.
Claude became my coding tutor, walking me through everything from setting up GitHub, and later Replit. It gave me step by step instructions on how to creating forms by integrating Formspree - also helpful for handling submissions. It was like having a patient developer (with high EQ) sitting next to me, explaining not just what to do, but why each piece mattered.
I deployed the whole thing on Replit, which turned out to be perfect for rapid prototyping and testing. The platform handles the infrastructure headaches so I could focus on building tools and give it dumb prompts like - “make it sleek” or “do you think someone would answer that?”
Mind Blown
The moment everything clicked was when I realized how easy it can be to use AI to build something that could actually bridge the gap between relationship science and real-world matchmaking. Professional matchmakers have always relied on intuition and experience, but now they could also leverage decades of psychological research through AI.
What really blew my mind was how natural it felt to translate psychological research into practical matchmaking tools. The Big Five personality traits, attachment theory, relationship psychology – all of these academic concepts suddenly became actionable insights that matchmakers could use to help their clients.
Here are some screenshots of the app -
First landing page for the website.
Matchmaker view
A Client Intake form:
Compatibility Matches based on input (not real people :))
What I Learnt
Building Intention taught me that the most important technical skill isn't coding – it's understanding human behavior. And the fact that artificial intelligence is basically a HUGE enabler for creativity and real-world problem solving.
If you’re a lawyer reading this, trust me when I say the legal profession has prepared you for this in ways you may not have expected. Lawyers are professional problem-solvers, and that mindset translates beautifully to product development. We're trained to think systematically, anticipate problems before they occur, and build solutions that work under real-world conditions.
I also learned that imposter syndrome is real, but it's also overrated. I'm not a professional developer, and I'll never write code as elegantly as someone who's been doing it for years. But I understand my users because I am my user. I think with the advent of generative AI, that intimate knowledge of the problem space matters more than perfect technical execution.
Most importantly, I learned that the best way to learn is by doing. The app is obviously just a prototype and in no way perfect. I’m not sure what I will do with this idea even. But the fact that I was able to think it and build something, anything in a couple of hours, feels like.. magic.
Intention is live at web-socket-chat-ninjalawyer1.replit.app . Because what's the point of building something cool if you can't share it with the world? Please do not use it for actual matchmaking - I do not want your personal info, though I wish you the best as you look for love. :)
If you enjoyed this post, subscribe to Arguments with Algorithms for more stories about what happens when lawyers learn to code.