The conversation below is part of an ongoing series with founders about what they’re building and why it matters. It has been edited for length and clarity.
Algorithms have an answer for everything except what's good. They can tell you what's trending, what people like you clicked on, what to watch next — but they can't tell you why something matters, or what it felt like to discover it. That gap is widening. As AI-generated content fills every feed and recommendation engines collapse taste into pattern-matching, the most valuable signal online is increasingly the simplest one: a real person, with a real point of view, telling you about something they love.
Tyler Bainbridge has been building around this idea since 2020. He describes his platform Perfectly Imperfect (one of our portfolio companies) as "the world's largest taste anthropology project": 800,000 recommendations and counting, all human-curated, none algorithmically sorted. Contributors range from Francis Ford Coppola to your best friend's cool older sister.
Below, Tyler talks about why status-signaling is actually a feature, what makes a recommendation worth trusting, and why the best discovery engine is still just a person with good taste and a reason to share it.
At scale, doesn’t taste just collapse into sameness or status-signaling?
I actually believe that status-signaling (when it comes to recommendations) is a good thing. When someone shares something niche that you’ve never heard of before, the correct reaction should be “wow, I should check this out” rather than “ugh, that’s pretentious.” Taste will always function as social currency (whether it’s intentional or not) and to build a reputation as a tastemaker - you have to share. A human powered discovery flywheel.
As the PI platform scales we want to guide you towards human curators (not content) that you’ll like. By following them you’ll see other facets of their interests too (movies, products, etc). This is by design and helps you expand your own taste.
Is a Perfectly Imperfect recommendation meant to last, or is the value of your archive actually found in its ability to capture a vibe that is meant to expire
These days I dress differently and listen to other music than I did 5 years ago, but there’s someone out there who is yet to discover Online Ceramics or JPEGMafia — and those old recs might bring them new joy.
I look at our data set of 800,000 recommendations and see specific eras of taste. For instance, 2025 was massively dominated by people posting about Geese and Cameron Winter, and our interviews from 2021-2023 documented NYC’s art scene post-COVID. Cool might fade and morph, but documentation and context allows it to persist forever.
Perfectly Imperfect is the world’s largest taste anthropology project.
What makes for a great recommendation? Where do people get it wrong?
A great recommendation that’s shared without personal context, is ultimately worthless. I need to know who you are, why you like this thing, and what else you’re into, in order to make an informed decision on “is this worth my time.” There’s far more weight to a recommendation if it comes from Lou Reed (i.e. his Yeezus review for Talk House, one of the all time best pieces of music writing, ever) than a co-worker, a friend’s recommendation is more valuable than a stranger, and so on.
I got introduced to someone when I first moved to NYC, and our mutual friend said “you guys should be friends, you both like Jonathan Richman” which led to us talking about music for hours. Fast forward five years, and they’re now one of my best friends and we send each other songs all the time.
That’s the feeling I want to capture, and scale, with PI. Taking this basic human framework, and building it into a network of tastemakers. Everyone deserves to have friends (online or IRL) that they can nerd out with.
PI is often described as an "old internet" experience. What role do aesthetics play in how people behave on the platform?
We designed our app to be maximal and rough around the edges as a callback to when websites had personality. But I don't think this is purely nostalgia — it's more profound than that. Individuality has been stripped from social and we're now in an era of consumption vs. sharing. People are posting less and mostly consuming viral reels, DM-ing them to friends, and doomscrolling. We believe you'll share more of yourself if you're given a corner of the internet that's customized by you. And that's proven true — our d30 retention skyrockets to 85% if a user themes their profile or shares a few recommendations.
Does culture require a certain level of gatekeeping to remain culture, or is it purely a democratic measurement of attention?
We feature people who shape culture - whether it’s top-down from A-list celebrities, or bottom-up from underground artists/creators who are shaping the future. The latter is fundamentally anti-gatekeeping. We don’t want our favorite up-and-coming artists to remain niche for selfish reasons. Instead, we’re screaming from the rooftops “pay attention! this person is doing something important!” in the hopes that it shifts macro culture in some way.
What would you be doing if you didn't work in your current job?
I’ve been secretly hacking away at a movie script in my spare time and I’m confident my film will be made someday! Perfectly Imperfect Studios has a ring to it…
What's something that's overlooked by others but you can't stop talking about?
A few years ago I got into bowling with a few friends and now I have 3 balls, bowl twice a week, and it’s one of my favorite things to do. People talk about how we need more third spaces, etc, and I think the slow death of bowling has had a lot to do with this. It used to be much more common for friends, family, and strangers, to gather over a pitcher of beer and spend an hour on the lanes — it’s crucial americana!
What's an underappreciated corner of the internet?
RateYourMusic - it’s looked basically the same for the entire 20 years I’ve been using it, but if you’re a music nerd this is basically heaven on earth. Decades of user submitted reviews, artist catalogs, and charts. As a teenager in the suburbs without a cool older brother type figure, I needed this.
What do you wish you had more time for?
Bikepacking - I’ve always wanted to bike across the United States, but as a busy founder and husband it’s nearly impossible to disappear for 30 days and live on the road. Someday!
You can learn more about Perfectly Imperfect here and find Tyler’s profile.
The bit about personal context making or breaking a recommendation is the real core here. Spotify knows my listening history better than any friend does, but a friend's rec carries something the algorithm never will - the implicit "I know you, and I think this will mean something to you." That's not a data problem. It's a relationship problem, and no amount of collaborative filtering gets you there.
The concept is great. The execution is terrible in my opinion. It appears to be unusable unless you invest a lot of time in trying to figure out how it works. Maybe I’m just too old being a non digital native.