john sillings dot com

Thoughts on the current (un)profitability of the AI ecosystem

2025-10-22T00:00:00+00:00

Folks have been claiming that we’re in an AI bubble for the past 18 months or so (hand waving). Given enough time, the claim that we’re in a bubble usually turns out to be right.

But the more interesting claim is that AI is a house of cards – basically unprofitable the whole way down – and we’re due for a massive unwinding.

There is almost certainly some truth to this, but there is a funny implication that the participants are irrationally setting themselves up for failure. I don’t think that’s the case at all.

We’ve seen this before

We’ve seen this sort of thing play out a few times now: in ridesharing, content streaming, semiconductor manufacturing, cloud computing, and other industries.

In each case, there’s a huge opportunity with economies of scale. Investors fund it aggressively, and companies build aggressively to capture share (unprofitably).

Then, there’s an unwinding – but it’s not always a crisis. Prices go up, but input costs also go down. Many companies die along the way. Some companies capture big market shares and become the new incumbents.

Input costs are the boring part of the story

The input costs going down part doesn’t get enough attention… probably because it’s the less explosive, dramatic story.

But there are lots of forces at work here: cheaper models, better model orchestration, more efficient GPUs, training amortization, and more. All of these make the (admittedly hairy) unit economics picture look a little better.

But it still could be bad?

It’s certainly possible – but we don’t need to freak out just because an big, growing industry is unprofitable. That is fairly normal.

The most likely outcome, as I see it, is a relatively boring story that won’t make headlines:

Prices will go up some
Input costs will go down some
Some companies will capture huge share along the way

Thoughts on using a computer efficiently

2025-09-12T00:00:00+00:00

Tl;dr: ditch the mouse, it’s for n00bs. Use keyboard shortcuts for everything.

If you’re spending every day on a computer, and especially if you do knowledge work, and especially if your brain is still relatively plastic… you should be using the mouse as minimally as possible.

Brain-computer latency

Moving and clicking the mouse is slow and imprecise. Keyboard inputs are fast, perfectly precise, and chain-able.

Speed is awesome, and it adds up.

But the real value is in reducing the latency between your brain (“I want to do this thing on the computer”) and the computer input (“I did it”).

If you’re doing deep work on a computer, keeping this latency low helps you get into a flow state. When the line between having an idea and executing on it disappears, speed and quality both skyrocket.

(And if it isn’t intuitive how valuable this is: look up videos of Excel world champions – and then compare that to how you use Excel.)

Be a power user

Every program worth its salt – web browsers, spreadsheets, IDEs, video editing software, DAWs – has a crapload of keyboard shortcuts for power users. If you use a program frequently, you should learn them.

OK, fine – but how?

Here’s some really great advice my old boss, Dan Ha, gave me, paraphrased:

Every time you find yourself doing something more than once, look up the keyboard shortcut and use it. Then just stay disciplined, and make sure you keep using it.

It’ll make your work faster, better, and more fun: I guarantee it.

Thoughts on how to disagree

2025-09-11T00:00:00+00:00

Back in the day, I was on my school’s debate team.

It was basically a sport for nerds. (In hindsight, my friend group did a lot of “sports for nerds” activities: chess, quiz bowl, competitive gaming, etc.)

Anyway, back to the debate team. We treated it like a sport. We trained for it and attended competitions. It was a lot of fun – but deep down, we were there to win.

A stronger, better-researched, more clearly articulated argument was the big key to winning (obvious).

But there were lots of other ways to score points:

by flustering your opponent with a hard question,
catching them off-guard,
pointing out something silly they said,
and so on.

This is how most people think about debating, whether it’s on social media or between political candidates. It is eristic (about “winning”).

Obviously, this is a popular form of entertainment. We enjoy the drama. We like seeing other people (particularly those we disagree with) get “owned” / “dunked on” / “destroyed” / etc.

What sucks about this is how shallow it is.

Mostly in academic circles, there is a less popular and considerably less entertaining form of debate. It is dialectic (about learning through disagreement).

The ground-rules are different:

You are charitable. You interpret your opponent’s argument in a strong way, not a weak way.
You look for common ground. If you find yourself agreeing on something – however small – you build from there.
You put your ego aside. Being wrong is not a bad thing; it means you’re learning something new.

In a narrow, self-serving sense, there’s obviously a lot of value in being persuasive. But in aggregate, it’s a lot more exciting to actually be right vs. simply being perceived that way.

This post, and particularly its title, was inspired by this excellent essay by Paul Graham.

How far away is a fully AI-run business?

2024-04-15T00:00:00+00:00

A demo video featuring an AI programmer called Devin went viral recently. The video depicts an AI agent that takes instructions, formulates a plan, writes code, fixes bugs, and iterates its way toward a shippable solution.

Even more recently, evidence came to light that this demo was staged – and that the actual capabilities of Devin are a lot less impressive than depicted.

I’m a lousy programmer who uses AI all the time to hack away on small projects, build prototypes, and write useful scripts. Back in the day, we used to call people like me “script kiddies.”

But I know people who are actually skilled programmers – friends and colleagues – who are also finding a ton of value from LLMs like GPT-4 and Claude for coding.

So even though the internet is having fun with another oversold, overhyped AI product, I’m actually quite bullish on AI agents like Devin.

In fact, I’m working with one such company right now – Checksum. Checksum is much more specialized than Devin: it writes and maintains end-to-end tests for web apps. But even in these early paces, Checksum’s orchestra of models does its job extremely well.

More importantly, the pace of progress has been staggering and doesn’t show signs of slowing. I’m lucky enough to see this from the inside at Checksum and other businesses I work with – but I think it’s pretty obvious to any outsiders who are paying attention, too.

Building software is an obviously valuable target for an AI agent. But other core business functions, like customer acquisition, customer support, and general purpose activities like browsing the internet and sending emails are also getting the AI treatment.

So – how far away are we from a fully-AI run business?

It seems like we’re pretty close.

Can an AI agent build and maintain a simple software product? Years away.

Can an AI agent do customer support? Already here.

Can an AI agent browse the internet? Already here.

Can an AI agent pay for stuff? I don’t see why not.

Can an AI read and respond to email? Already here.

Can an AI agent build and maintain customer acquisition channels? Years away.

(This one is particularly interesting to me. Digital channels that are text- or image-based will likely be first up.)

Can an AI agent take information from one system and give instructions to another system? Years away.

(The generalized LLM part of this is here, but the generalized action-model part may be years away.)

So what?

Emotionally, I’m not sure how I feel about all this.

But it sure as heck looks like it’s coming either way – and it’s been hard to think about anything else lately.

Validate with cold email

2024-03-13T00:00:00+00:00

I spend a lot of time thinking about the best way to validate startup ideas.

And having tried quite a few things, I really think cold email is undoubtedly the best way to validate your B2B startup idea.

It’s brutal

Cold email is brutal. It’s not as bad as cold calling, but it’s unpleasant enough that a lot of people outright refuse to do it.

If you email 100 people and get 1 response back, that’s considered decent performance in the unforgiving world of cold email. And if you have a poorly written or poorly targeted campaign, your annoyed recipients will certainly let you know.

But this weakness is also a strength. When you do get an interested response back, you know it’s genuine. You’ve already broken a social contract and barged into somebody’s inbox; they don’t owe you any politeness. They actually care. They’re actually interested.

Other channels (like events) and approaches (like talking to your friends and colleagues) are rife with social desirability bias. Entire books – including really good ones like The Mom Test – have been written to help you avoid this bias.

But you don’t need to be an expert in conducting user interviews to validate your idea. You just need to do the hard, slightly presumptuous thing, and bother some unwitting strangers with your cold pitch.

It’s great for experimentation

Cold email gives you an incredibly high degree of control over

who gets your message
what message they get

On easy mode, you can simply select what features you’re interested in from a database: job title, industry, location, funding, and so on.

But things get really interesting when you start daisy-chaining different systems together to put a really fine point on your targeting. Between scraping, the multitude of available databases, and LLMs acting as connective tissue, there’s virtually no limit to how precise your targeting can get.

And when it comes to your message: here, too, the only real limit to your experimentation is your own imagination. (And the technological constraints of email, which hasn’t changed all that much in the last 20 years.)

It’s measurable

There are good reasons to believe that cold email performance is actually pretty tough to measure. This has to do with the nature of spam filters, how various email clients work, and the oblique way that cold email providers guess about things like inbox placement and deliverability.

Even if you’re skeptical about metrics like open rates, you can always hang your hat on your interested reply rate. This is super straightforward to measure, and easy to compare from one campaign to the next.

It’s great for fast feedback

When it comes to email, the default behavior is typically:

respond relatively quickly (hours or days)
ignore forever (archive or just let it rot)

So another big point for cold email is the ability to get feedback very quickly. In a startup environment, this speed to validation is incredibly valuable.

It’s quick to get started

If you want to send a high volume of cold email, you’ll need to warm your domain up. This takes a little time, but not much – typically a few weeks.

But if you’re actually thoughtful about your targeting, you can move way faster.

Conventional wisdom is that you can send between 20 and 30 cold emails from your primary domain without risking your domain reputation (i.e. getting flagged as a spammer).

That’s plenty of volume to send some thoughtful, personalized emails and generate real interest… assuming you’re actually onto something.

You can generate real signal very, very quickly: typically within days.

It’s not perfect

There are a lot of reasons to not like cold email, and I don’t want to come off as a total fanboy.

It doesn’t always scale. It can be annoying to maintain. It’s a pain in the neck to troubleshoot.

On a fundamental level, it can feel like a trespass into someone’s personal digital space. It has the propensity to annoy. And if you do it enough, you will get cursed at, yelled at, and insulted.

Particularly when it comes to the social aspects of cold email, a lot of folks fear it. Most people don’t want to be annoying or rude.

But when it’s thoughtfully done, and when you’re grasping around in the dark at the early stages of your startup with your unvalidated idea, it can be an extremely clarifying and effective place to start.

40 weeks of (mostly) bootstrapper-friendly cooking projects

2023-10-13T00:00:00+00:00

For years and years, I pretty much never cooked.

When I was a kid, my parents did all the cooking and I wasn’t particularly interested in it.

When I was in college, I experimented a bit (mostly with the goal of bulking and consuming tons of protein), but just made the same few dishes over and over.

I also drank a full gallon of milk every day for a while, but that’s a story for another time.

After college, I also made the same few basics over and over again (like sandwiches, pasta, and bowls of cereal) – but I ate out a ton because it was convenient and I could afford to. Also, no place like NYC to eat out.

But then I quit that job to bootstrap not just one startup (Art in Res), but two (Mini Exhibitions).

I needed to eat cheap, but I still wanted to eat well. So I consumed – conservatively – thousands of YouTube and Tiktok cooking videos and taught myself to cook all the things I wanted to eat.

These days, I have a full-time job at a startup studio, Superset. But those bootstrapper-era constraints are still my guiding principles. For me, the best recipes are:

Cheap
Efficient (in terms of prep time and cleanup)
Reheatable and/or storeable with some shelf life
Modular
Tasty (obviously)

In an effort to keep improving my skills, and not fall into a “I’m good enough” cooking rut, I make a new dish every week.

Here are the last 40 or so. Maybe you’ll be inspired to try one or two out. I’ll also drop links and references throughout.

Chicken cracklings (ie; roasted chicken skins)

I’ll always peel and roast the skin from a rotisserie chicken from now on. This stuff is legit better than bacon. A link, but it’s really pretty basic.

Jar-shaken butter

Honestly, not really worth it; but could be useful in a pinch (especially if you keep heavy cream on hand like I do).

Pressure cooker chicken stock

This stuff tastes so good, and it’s a good way to use up the mirepoix-y vegetables that are getting a little stale in the crisper drawer. I just used the carcass from a rotisserie chicken.

Dutch baby

This one gets big thumbs up from the adults and kids – and it’s fun for everyone to make together. I went by the NYTimes recipe.

Oreo-butter stuffed crepes

One of my kids demands crepes at least once a week. This week, we broke out the molcajete, smashed up some oreos with a little cream, and filled a few crepes with the resulting “cookie butter.” It was a frighteningly efficient way to consume Oreos.

Pumpkin spice syrup

My partner likes pumpkin spice coffee drinks but they’re like $6 a pop at the coffee shop, so I decided to make the syrup at home. Great flavor, but I struggled to get a good, non-sludgy consistency. Link if you want to make your own spiced pumpkin sludge.

Blueberry lemon pancakes

Usually I’ll just make some pancakes and throw blueberries in. But adding a little lemon to the batter really set these off.

Blueberry jam

We bought a copious amount of blueberries at the farmers market, so I turned them into jam once they started to turn.

Feta cheese foldovers

This was my first time working with puff pastry, which is kind of a pain in the ass. But these were pretty good – I topped with honey and sesame seeds. Probably won’t make again, though.

Baked shrimp

Maybe I’m doing something wrong, but baked shrimp kinda sucks. They were sort of rubbery and soft, so I think I’ll stick to pan frying. In general, I think shrimp are more trouble than they’re worth, and kind of pricey too.

Baked boneless skinless chicken tenders

The kids often want “plain chicken,” and the adults can easily zhuzh this up for salads, wraps, sandwiches, etc. This is so far the best way I’ve found to bulk prepare it.

Homemade milkshakes

I used the Adam Ragusea tutorial, which is more of a format than anything. These were amazing and super easy to make with an immersion blender.

Garlicky sauteed spinach and leek

We always have some about-to-go-bad spinach in the fridge, but I find regular spinach to be a stringy globby mess that I choke down for health reasons. Adding some leek did wonders for the texture and added some nice allium-ness to the equation (along with the garlic).

Cocoa Puff marshmallow treats

This was a late night craving, and tasted good, but kind of looked like spider eggs.

Patty melts

Patty melts > cheeseburgers. And I love cheeseburgers. This was a certified slapper. I loosely followed Sam the Cooking Guy’s recipe.

Mango toast

One of our nieces recommended we try this, and it was pretty good! Basically buttered toast with mango on top; sort of a chunky jammy toast vibe. I think I prefer mango on its own, though.

Chicken tikka

I tried my hand at our go-to Indian food order, and it was absolutely delicious. I was kind of pessimistic that it’d stand up to our go-to restaurant’s version, but it actually blew it out of the water. A lot of work and cleanup, though, or I’d make it all the time.

I used Joshua Weissman’s recipe.

Papad (various)

One of my work friends, Pooja, sent me the papad that she makes with her family – all the way from Kerala in India. And what a care package it was: 10 different varieties, each tastier than the last. Fun to cook, too.

Dark ‘n’ stormy

My parents brought back some rum from Bermuda, and some ginger beer to go with. Never had one of these before, but they’re excellent.

Truffled scrambled eggs

We have a newborn who is not tolerating dairy proteins, so I’ve been trying to whip up some dairy-free – but still creamy and delicious – dishes for my breastfeeding partner to enjoy. The inclusion of truffle oil turned an otherwise ordinary egg dish into something pretty special.

Beets

Just straight-up beets. Never cooked ‘em before. I did ‘em in the oven with olive oil, salt, and pepper. They’re pretty good!

Blistered green beans with crispy chili oil

I always get these when I see ‘em on a Chinese food menu, so I decided to make ‘em myself. They were deceptively labor-intensive for such a simple looking dish, but they tasted great. I followed this recipe.

No-bake peanut butter chocolate bars

Made these to satisfy a craving. Obviously they were delicious, but I just felt like trash afterwards.

Chicken caesar salad wraps

Sort of a joyless meal to cook and consume, but definitely convenient and reasonably healthy.

Chocolate ganache

Dead easy to make in the microwave, and the kids are always begging me for chocolate-covered treats (e.g. strawberries, pretzels), so this was a good one for the arsenal. Shari Mukherjee taught me how to make this.

Irish soda bread

I dream about this bread from the couple trips I took to Ireland, and it was actually not too difficult to make something comparable. I’m the only one in the family who likes it, though, so I probably won’t make it too often. This was another Shari Mukherjee recipe.

Tomato confit

For a number of reasons, this is now a go-to for us. It tastes great, it’s (I think) reasonably healthy, it’s super easy to make, and my partner’s mom is always giving us heaps of beautiful cherry tomatoes that she grows.

Tomato sandwich

The farmers market near us sells these huge, juicy heirloom tomatoes that make for a great sandwich. This is also a new go-to because it’s easy, refreshing, and super satisfying. My friend Eric Kim inspired me to try these – they’re in his cookbook.

Bagel chips

Another farmers market thing – I buy lots of bagels at a time which I occasionally forget to freeze. But like making croutons, this is a really simple way to take stale bread and make something a little more shelf-stable.

Sushi rice

Sushi rice is so insanely good and totally worth the little bit of extra work to season it. Also the techniques of mixing and cooling the rice (which I suck at, but whatever) are really satisfying and fun.

Salmon rice bowls

Sometimes I have these without sushi rice, and it’s serviceable that way, but you really need the tartness of the rice vinegar and kick of sweetness. This is now a regular part of the rotation and we mix-and-match the toppings based on what we have lying around, so points for modularity.

White bread (in the breadmaker)

My parents found our old breadmaker – complete with the original sticker showing it passed a Quality Assurance inspection in 1993. The thing still churns out great loaves of bread and our kids love it.

Milk bread (in the breadmaker)

We also made “milk bread” with milk powder, which felt like an interesting science experiment, but didn’t taste so different from the regular white loaf. Making a proper milk bread is on my list to try sometime.

Cinnamon-tossed apple slices with peanut butter

When I went to my childhood friend David’s house, his mom would always serve us apples with peanut butter. I tried tossing the apple slices in cinnamon and it was so good that this is my new favorite way to enjoy this nostalgic favorite.

Spicy honey syrup (for cocktails)

This was an unwitting collaboration between the parents (mine and my partners) – hot peppers from her mom, and honey from my beekeeping dad. This stuff was awesome; I had a little bit but gave most of it away as a gift to a mixologist friend.

Lazy pizza with grocery-store dough

I never actually used the grocery store pizza dough before, but it actually makes a pretty tasty (if sort of bready) pie. It’s easy enough that the kids can basically do the whole thing.

Chicken riggies

This is a Utica, NY specialty that’s somewhere between penne alla vodka and chicken tikka masala. It’s super good. My partner showed me how, but here’s a recipe to check out.

Quick Detroit style pizza

This is my go-to pizza recipe – but I think you really need one of the Lloyd pans (or something similar) for it to come out great. You can buy one here on Amazon – not cheap at $60, but worth every penny.

I use Ethan Chlebowski’s recipe.

Pickled red onions

Speaking of Ethan Chlebowski, I made pickled onions the very slightly more labor-intensive way (not the lime juice quick-pickle) and it was 100% worth the extra 5-10 minutes. Recipe.

Mega onion focaccia

I am an onion and allium lover through and through, but this was even a little much for me. I caramelized three types of onions and absolutely slathered this thing with them. It was still pretty damn good, but needed some other complementary flavors in the picture.

I use Joshua Weissman’s foccacia recipe which is just next-level.

Mushroom risotto

Despite how labor-intensive this dish is, I ended up making it a bunch more times because it’s just so good. I wish there was an easier way… Also, pretty neat that you can get such a creamy dish without milk – I always figured there was some milk (or cream)! Chef Tracey showed me the way.

I usually write about startups and tech, not food. But if you want to read an essay I wrote about tortillas, go for it.

I'm already nostalgic for AI-generated Seinfeld

2023-02-05T00:00:00+00:00

It’s been a wild year for artificial intelligence and machine learning, and Nothing Forever is easily one of my favorite things to come out of it.

If you haven’t seen it yet, the premise is “AI-generated Seinfeld,” streaming 24/7 on Twitch.

The dialogue is written by GPT-3, and according to the creator, nearly everything else about the show is generative, including the speech, camera cuts, shot length, scene length, character blocking and movement, and music.

On its own, Nothing Forever is pretty amusing and charming. But with Twitch chat streaming alongside it, it’s hilarious, chaotic, and gives me this weird feeling of anticipatory nostalgia.

I don’t know what the future of AI will be, but I am confident that it will continue to evolve rapidly in the near future. (For what it’s worth, I am bullish on AI.)

The current state of AI and ML – and weird projects like Nothing Forever – reminds me a lot of the early consumer internet. I’m talking dial-up AOL, Netscape browser, Geocities, Hamster Dance, Flash animations: that internet.

Both now and then: it’s so exciting and everything is changing so fast. Even though things are rough around the edges, that’s part of its charm. The possibilities feel totally vast and totally underexplored.

One thing that feels particularly prominent this time around is self-awareness around how janky things are, and that’s really the thing that I love about Nothing Forever.

The dialogue is bad, and Twitch chat is a chorus of laughter and celebration when the AI produces “an actual joke!”.

The characters’ speech is stilted and awkward, and chat revels in mocking Larry, Fred, Yvonne, and Kakler’s weird intonations and emphases.

Characters walk through furniture, fuse into one another, forget who said what, and randomly go silent – and it’s all great fun.

For a certain type of person who likes technology and absurd humor, it’s an awesome cultural moment – one I’m happy to have witnessed. And if it’s not obvious enough, this is a ringing endorsement to check it out.

At the very least, it will go down as a great meme – but it might just foreshadow something about the future of media, too.

Breaking the behavioral seal

2021-12-05T00:00:00+00:00

I’m not a self-help guru.

In fact, by any standard measure, my life is pretty chaotic.

But this little mental model, “breaking the behavioral seal,” has been fantastic for my personal and career growth, so I’m sharing it with you.

My first time making tortillas

A few years ago, a close friend taught me how to make flour tortillas from scratch. I love Mexican food, and I’d been wanting to try it for a while – years, in fact. But somehow I never got around to it.

So, my friend had me over and graciously taught me: this is the ingredients list, the prep, the timing, the technique.

You don’t need much to make flour tortillas: flour, fat, warm water, baking powder, and salt. A mixing bowl is a must, and a rolling pin or tortilla press are nice to have, but you can improvise.

They come together fast, and even “bad” homemade tortillas blow the grocery store ones out of the water. But the first batch came out great, and were quickly consumed.

A few days later

I’m home, hungry, and tortilla-less.

And despite having almost no experience and no rolling pin, I got to work. And they came out … pretty bad. But I enjoyed them all the same.

So after years of merely thinking about making tortillas, I made them twice in the span of a few days.

And then, about a week later, I made them again. And the following week, again.

These days, unless I’m traveling, I’ll make a batch every week or two. It’s a little ceremony I cherish.

And it all started with just … starting.

“The first step is the half of it.”

I’m not the first person to happen upon this insight. Another close friend (who has eaten many of my tortillas, by the way) shared this Korean proverb with me: “The first step is the half of it.”

Put less elegantly but more descriptively: if you have a goal, you stand to make a disproportionate amount of progress by simply getting started.

And more uncouth: once you break the seal, there’s a decent chance you’ll keep going.

People who break the behavioral seal

I’m lucky to know some extremely effective, skilled people, and they all make a habit of breaking the behavioral seal. What makes them try new things?

They have an action bias. They don’t just idly talk or think about doing something, they actually do it.
They love learning, and are unafraid of (and even excited by) being a novice.
They are resourceful. Even if they don’t have much knowledge, the right tools, or a great environment, they make it work (or try to).

And people who don’t

They make the same handful of excuses:

“I’ll do it later.”
“I already know I won’t be good at it.”
“I don’t have the right tools or the right environment.”
“I don’t know what I’m doing.”

Getting over the hump

We all have thoughts and conversations with friends and colleagues that go like,

“[I / we] should do [a thing], that would be [fun / good for business / interesting / healthy / a great experience].”

And we all know how this story typically ends: talk about it, indulge the fantasy, and never do it.

When we could simply: do it.

Sure, it might be hard. Sure, you might fail. So what?

Stop fantasizing about making good tortillas and start making bad ones. It’s the best way to make good ones.

Mistakes you're probably making in the Y Combinator interview

2021-11-30T00:00:00+00:00

You got invited to interview with Y Combinator – time to celebrate a little and pat yourself on the back! Or is it?

Getting to this stage is an accomplishment, but it isn’t time to declare victory. It’s time to buckle down, prepare, and ace the interview.

These are the top mistakes you’re probably making, and suggestions for how to fix them.

The main problems

You’re talking too much: just answer the damn question

The YC interview is 10 minutes long. That is short.

When a partner asks you a question, just answer it directly and concisely. If they want more information, they’ll ask.

Suppose you get asked: “What’s your churn rate?”
It’s great to say: “It’s 5%.” If the partner needs more information, believe me: they’ll ask.
Don’t say: “It’s 5%, but we think it’s going to go even lower once we implement this new feature that we’re working on. We talked to one of our top customers and…”

You’re smart, you’re excited about your product, and you could talk about it all day long. But now’s not the time. Answer the question directly.

Answering concisely makes you look better, and helps the partners get through all their questions.

You don’t actually know your numbers

Say it with me: you need to know your numbers.

And this isn’t just an interview thing. This is a founder thing. This is a business operator thing.

If you don’t know your numbers, you don’t really know what’s going on in your business.

Table stakes: Revenue, growth, costs, profit, retention rates, churn rates, average sale prices, active users, website traffic, burn rate, conversion rates through your funnel, repeat buyer rate, customer acquisition cost, lifetime value per customer, total addressable market, industry growth rate.
Big brain: You know how those metrics have changed over time. You know the metrics for different subgroups (e.g. different customer types). You’re aware of industry averages and competitor metrics. You can talk about specific customers (and, if relevant, suppliers) in a numerical way.
Don’t do: Uhhh, uhhh, uhhh…

Your intro sucks and/or you don’t have it memorized

You need to have your intro down cold.

You intro should consist of who you are, what you’re working on, and what problem you’re solving. The goal is to be descriptive, but also a little exciting and thought-provoking.

Our intro was something along the lines of: “We’re Art in Res, an art rental marketplace for consumers. We let you try art before you buy it, and buy it off your own wall in installments.”

After we got in, we updated it: “We’re Art in Res. We lower the entry barriers to buying fine art by letting people try it before they buy it. Instead of paying upfront, you can buy art in installments and can return any time.”

We pivoted Art in Res during the W20 batch, but that’s what we did at the time.

Do: Introduce yourself, explain what your company does and why it matters.
Don’t: Ramble on after your intro, or wing the intro.

Who gives the best feedback?

2021-11-27T00:00:00+00:00

Are you building something or making something new?

Well, you don’t want to build in the dark. You want to make sure you’re building something people want. In order to do that, you need to talk to the right people.

The how-to bible of collecting user feedback is undoubtedly The Mom Test. If you’re serious about doing user interviews the right way, this is a must-read.

Let’s assume you’ve done your basic homework and you’re ready to start getting feedback on your idea or your product. Who do you talk to?

Here’s my semi-serious, biased, and heavily anecdotal ranking of the different types of people who will give you feedback.

The different people who will give you feedback, ranked

They’re the best. They don’t care about hurting your feelings and won’t sugarcoat when it comes to discussing their issues. They actually care about solving their problem and that’s why they’re speaking with you.

Normal people who actually use your product

They will sugarcoat some things. But they’ll be generous with their time, give you real feedback, and tell you what they love about your product so you can double down on what’s working.

Friends of friends who fit your ideal customer profile, talking behind your back

Great for impressionistic feedback. They’ll speak candidly and truthfully about big-picture things like your branding and feature set (but not to you). A great opportunity to learn about why more people don’t use your product.

Very smart investors and domain experts

They have an opinion about you before they even talk to you, and it may be right. Their interests are not necessarily aligned with yours. But, they’re very smart and probably have way more perspective than you do, so it’s an opportunity to learn some super valuable lessons.

People who almost used your product, but then didn’t

One of the best people to talk to. Usually not generous with their time, and very hard to pin down. If you can find out why they didn’t use your product, that’s a huge win.

People who used your product and churned out

Same as the persona above: hard to pin down and usually not generous with their time. But if you can figure out why they stopped using your product, that’s extremely valuable.

They don’t care so much about hurting your feelings, so they might give you useful, honest feedback. They will be generous with their time. But they’re probably not your target customer.

Operators in your industry who aren’t competitors

There is a good chance they will sugarcoat their feedback, which is hypothetical anyway. They may have an ulterior motive. But they will have unique insights which are extremely valuable, and might share them with you.

Reasonably smart investors

They have an opinion before they even talk to you, and it’s probably flawed, if not objectively wrong. But they have a lot of perspective and experience, so you may learn a valuable lesson or two.

Friends and family

They want you to be happy, so their feedback will be biased. They will be extremely generous with their time.

Anonymous haters on the Internet

Their primary goal is to signal their intelligence in a public forum for internet clout. You may learn valuable impressionistic things from them, but you may be misled, too.

Anonymous nice people on the Internet

They want to be nice and helpful, but they’re also chasing internet clout in a public forum. You might learn something useful, but you usually won’t.

Direct competitors

Highly dependent on the person and how “zero sum” their attitude is. They will have an extremely valuable perspective, but they will usually be cautious about sharing information with you.

Inexperienced smart people who think they know what’s best for you and/or your company

Don’t bother. Extremely difficult to extract unbiased feedback. In the supportive case, they want to tell you what to do and how to do it. At best, misleading. At worst, demoralizing.

Our Y Combinator interview prep questions

2021-11-26T00:00:00+00:00

Are you prepping for the YC interview? Here are the questions we used to prep for our interview. We applied with our online art marketplace (formerly art rental marketplace) Art in Res and got in.

H/t to John Friel, my Art in Res and Mini Exhibitions cofounder, who thought up and compiled these questions with me.

A few general tips:

Answer every question as precisely and succinctly as you can.

Avoid using jargon.

It’s fine for one person to do most of the talking, and the interview may go more smoothly this way.

If a cofounder is doing less talking, decide pre-interview which questions (or question topics) they’ll be responsible for answering.

The questions below are general questions – you should also anticipate lots of domain-specific and product-specific questions.

Take care to think about possible tough questions and possible objections pre-interview so you’ll be prepared. You don’t want to get flustered.

And now, on with the questions.

Product and Product Demand

What are you making?
How does your product work in more detail?
Who needs what you’re making?
How do you know customers need what you’re making?
How will customers and/or users find out about you?
Where do your new users come from?
What makes new users try you?
Why do the reluctant users hold back?
Where is the rocket science here?
What is the next step in product evolution?
What’s new about what you make?
What are the top things users want?
What has surprised you about user behaviour?
How are you understanding customer needs?
Who were your first customers?
How are you meeting customers?

User Behavior

How many users are paying?
Are your customers one-and-done, or do they come back?
What are other ways people solve this problem?

Progress

What are your numbers? (to be elaborated into more questions)
Why haven’t you made more progress?

Business Model

How will you make money?
Why is your product priced the way it is?

Next Steps

Why do you want to be part of YC?
What are you going to do next?
Would you relocate to Silicon Valley? (maybe not relevant anymore)
How long can you go before funding?
Will your team stick at this?
How do we know your team will stick together?
Are you open to changing your idea?

Opportunity

If your startup succeeds, what additional areas might you be able to expand into?
How do you think about your ASP (average selling price) over time? Do you think this will go up, down, or stay about the same? Why?
What could you do to expand your TAM?
How big is this market, really?
How do you know that this isn’t just for [subculture or niche user base]?

Challenges & Obstacles

What’s the worst thing that happened?
What’s the biggest mistake you have made?
What problems / hurdles are you anticipating? How will you overcome them?
What will be the biggest challenge two years from now?

Competition

Who are your competitors?
Who do you fear the most?
What do you understand that others don’t?
Why isn’t someone already doing this?
How is your product different?
What, exactly, makes you different from existing options?
Who might become competitors?
What do you understand about your users?
What are the key things about your field that outsiders don’t understand?
Won’t [big company] just launch a [whatever your company does] feature and kill you?

Company History

Why did you pick this idea to work on?
Have you raised funding?
What’s the funniest thing that has happened to you?
Tell us something surprising you have done?
What’s an impressive thing you’ve done?
What systems have you hacked?

Company Formation

Will you reincorporate as a Delaware C-Corp?

Marketplace Questions

What side of the marketplace is harder, demand or supply?
Is it harder to find suppliers to list [whatever good or service], or buyers to buy?
How do you plan to build the marketplace from scratch?
How would it work to expand geographically?
Why is your commission so high?
Why is your commission so low?

Team

How do you all know each other?
Who are the other founders? (If you have <10% ownership founders)
Why did your team get together?
How did your team meet?
Who in your team does what?
Who would be your next hire?
Who would you hire or how would you add to your team?
Who is “the boss”?
What domain expertise do you have?
In what ways are you resourceful?

Metrics and Financials

What is your growth like?
How many users do you have?
What is your user growth rate?
How many users have ever engaged?
What is your burn rate?
What’s the conversion rate?
What’s the repeat buyer rate?
What’s the churn rate?
How much does customer acquisition cost?
What is the average price of a [purchase, subscription, etc]?

PS: If you’re getting ready for your YC interview, I wrote about some common YC interview mistakes (and solutions) here.

Perfectionism and pragmatism

2021-11-25T00:00:00+00:00

There’s a meme about job interviews that goes like this: The interviewer will ask you, “What’s your biggest weakness?” And you respond, “Probably that I’m a perfectionist.” I’ve never been asked this question in an actual job interview, nor have I asked it. But, I do think there is some wisdom in there. Perfectionist tendencies can make you a great fit for some jobs, and an awful fit for others.

My own career so far has basically been two jobs.

Career path 1, investing on Wall Street, is a great fit for a perfectionist.

The stakes are very high ($10-100M+ might be riding on a single investment).
A slight edge can be hugely profitable.
A slight disadvantage can be hugely punitive.
You can focus your attention on a particular area where you can outperform others (and you’re typically encouraged to).
You can ignore things that aren’t relevant to your investments and prospective investments.

Career path 2, tech founder, is a terrible fit for a perfectionist.

You have a huge list of to-do’s that feel important, and no choice but to triage.
Most mistakes can be fixed.
Most advantages last only as long as you keep investing in them.
In the early days, you and your cofounders need to wear all the hats (tech, sales, marketing, accounting, legal, HR, management, etc). And in the beginning, you’ll suck at almost all of them.
In most cases, quickly shipping and iterating beats trying to ship something flawless.

Is there something generalizable in here? I think so.

If you’re thinking about a job where small mistakes are particularly costly (civil engineer, aerospace engineer, surgeon) or small advantages are particularly well-rewarded (lawyer, investor), these will play to your perfectionist tendencies.

Or, are you interested in a job where you can iterate quickly (software engineer, marketer, writer) or your ability to ship is more important than avoiding mistakes (artist, entrepreneur)? Pragmatists will outperform.

As a first-time founder, it took time to unlearn the perfectionist tendencies I picked up as an investor. And despite all the reading and all of the good advice I consumed, I still made a lot of classic newbie founder mistakes as a result: I shipped too slowly. I expected too much from version one. I didn’t course-correct fast enough. I was too married to my original vision.

New job seekers and career-changers, I hope you can introspect on your own perfectionist or pragmatist tendencies and find the right career fit for you. And to the perfectionist founders reading this, take it easy on yourself.

What makes Michael Seibel special?

2021-10-23T00:00:00+00:00

If you went through Y Combinator any time in the past 7 years or so, you know: office hours with Michael Seibel just hit different.

And by saying this, I am in no way trying to downplay how fantastic the other YC partners are. There is obviously an incredibly high bar.

When you’re going through YC, or you have the opportunity to be advised by any YC partner, you can expect certain things:

First, if you’re prepared with questions, you’ll get a lot of questions back. You will be expected to have answers.

Then, you’ll get advice that may very well transform your company. The advice will, almost always, be direct and unambiguous – but sometimes the answer will be more like a heuristic.

It will feel like drinking from a fire hose, and by that I mean it will be fast-paced and you’ll cover a lot, quickly.

You’ll leave with a clear set of goals, a sense of urgency, and feeling of hopefulness and excitement.

But with Michael, it really is different. If you know, you definitely know – and if you don’t, I’ll do my best to convey what makes a conversation with Michael Seibel so inspiring and effective:

High intensity

In normal social situations, from what I’ve experienced, Michael is a very chill dude with a great laugh who makes others feel included and at ease.

But in the context of being advised by Michael: no. You’ll get the great laugh from time to time, but that’s about it.

We went through YC’s W20 batch (“the COVID batch”), the last in-person batch at the time of writing. I sometimes wonder whether remote founders in the more recent batches experience Michael’s intensity in the same way.

To say Michael “grills” you would be an understatement. The moment a question leaves Michael’s lips, the clock starts ticking. And less than two seconds later, you already can feel it and see it all over Michael’s face: “Does this idiot really not know how many unique visitors he’s had in the past two weeks?”

And, to add to the pressure (and this is where I wonder about the remote YC experience), Michael’s saccades – the movement of his eyes between fixation points – are on another level. You can actually feel the fixation on your eyes, the corners of your mouth, your shoulders, your forehead.

It’s one thing to be asked rapid-fire questions and be expected to have answers – that’s pretty standard. With Michael, it’s that, but on God mode. It’s that, but with a hypersensing supercomputer bearing down on you, wondering why you’re such a f***ing idiot.

I’m exaggerating a little bit, but that’s sort of what it feels like. And in case you’re interpreting this as a criticism, it’s not: this makes you a better founder. After a few conversations with Michael, you will hold yourself to a higher standard of knowing your shit.

Professional hater

If you follow Michael’s Twitch cofounder, Justin Kan, you might have heard Justin describe Michael as a “professional hater.” It’s kind of true. Michael is a straight savage.

Another YC partner might tell you that your landing page isn’t good and that you need to make some changes right away. Michael will tell you that your landing page is punching him in the face, and then he’ll ask you what the hell you were thinking.

This is another example of standard-raising, but it’s actually something more Socratic than that. “What the hell were you thinking?” is a really important question that sets the stage for a lesson you’ll never forget. With Michael, bad execution isn’t just fixed, it’s also scrutinized.

That anecdote about the landing page is a real one. We fixed the landing page – but we also had a productive conversation about what we were trying to accomplish in the first place. Later versions of our Art in Res landing page preserved the spirit of our original idea, but the execution was just way better.

When Michael goes into hater mode, the emotional stakes are high, but the commitment to understanding your decision-making process is high, too. The discrete ideas that went into your execution will be held up to the light and examined: “there’s something to this one,” maybe, but “this one is terrible.”

These conversations can be uncomfortable, especially to the uninitiated, but the lessons are both nuanced and enduring.

Chunking and pattern-matching

This one absolutely deserves mention, but feels more like a black box. I’m not a psychologist, but I promise there’s something to this.

You know how memory sport competitors (OK, maybe an obscure reference) use chunking to remember and retrieve huge amounts of information? Michael must be doing some version of that, because this man’s memory and pattern-matching abilities are absurd.

Pre-founder days, I worked in the investing world and I was a generalist. So I looked at all sorts of companies: healthcare, oil and gas, tech, telecom, insurance, whatever. Every industry and sub-industry was fair game, and every business model, too.

I’m definitely not a business analysis genius, but I’m good enough and experienced enough to know who’s good and who isn’t. All of the YC partners are excellent at this, but Michael is truly on another level.

The fluidity with which Michael can remember specific anecdotes and facts about companies he’s advised years ago, then transition from thinking about those companies as atomic units to groups of companies (with varying specificity across different dimensions), and do that with a sensitivity to time (the world changes quickly in Silicon Valley), while zooming in and out to varying levels of detail… it’s crazy to experience.

One time I remember this being on impressive display was early in our batch when we were thinking about pivoting our business. We got a call from Michael at about 8pm on a weekday. “What’s this I hear about you guys pivoting?”

The next 90 minutes were a tour de force. Michael analyzed at least 25 pivot ideas and problem spaces on the spot with zero preparation. Regardless of the topic, he could easily recall companies who succeeded, companies who failed, industry dynamics, strategies, regulations, products, pivots, growth channels, etc. But moreover, he would simultaneously theorize about possibilities: things that might work, things that probably wouldn’t, the direction of various industries, and the changing shape of the world.

There isn’t really a lesson here for founders, except that if you have Michael as a resource, you’re lucky. It’s evident that he puts in a ton of work and probably has some awesome genetic gifts, too.

Prescriptivist

This might be the one where Michael differs the most from the other YC partners.

I touched on this earlier, but YC office hours follow a general format. You come in with your questions, and you get asked questions back. You get great advice and the whole thing flies by. You leave and you usually feel pretty excited and good.

With Michael, though, there’s a decent chance that the meeting takes a turn. You come in with your questions, and you get asked questions back…

But then, something in the air changes, and you can feel it. Michael pieces something together. Your other questions are now irrelevant, at least for the time being. Michael is asking you more questions, and you’re at his mercy. The questions get increasingly psychological and emotional in nature.

And finally, the denouement. You get told exactly what you need to do. And you need to do it now, by the way. The floor is now open for questions, but except for a clarifying question here or there, you see how your other questions don’t really matter all that much. You’re inspired, you’re totally bought in, and you just want to get the hell out of there and back to work.

Everyone is fallible, and that includes Michael. In fact, Michael has said that about himself many times. Part of what makes startups so fun is the high degree of uncertainty and breadth of possibilities.

It seems perfectly reasonable for a smart advisor to remind themselves, “hey, I’m fallible, and this advice may be wrong.” The YC partners do this all the time, which is why it’s pretty common to get advice like “you should try X and Y” or “you should use heuristic Z to figure out the answer.”

Michael’s advice can often feel more like “And so it was told: Michael Seibel has spoken.” His ability to suppress his sense of fallibility and hit you with some blunt “do this”-type advice is sometimes exactly what you need.

Caring

I’m saving the most sentimental one for last, and I think PG’s essay about Ron Conway was sort of the inspiration for this essay. I wonder why it’s not more common to write essays about people who made your life better, or people you admire.

I’ve covered a bunch of ground. Michael will psychoanalyze you, obliterate you, teach you, and tell you what to do. But despite the high emotional stakes and uncomfortably blunt advice (that may or may not involve you feeling like an incompetent moron), I never for one second thought “Michael doesn’t care about us.”

In fact, quite the opposite – it’s pretty evident from all of our interactions with Michael that he cares a lot.

It would be totally fair for Michael – or any YC partner or investor, for that matter – to just care about their portfolio founders in the context of being an effective operator. But I truly feel like Michael cares about his YC founders in the way a teacher cares about their students.

He wants us to win, and he wants us to succeed, and it would be amazing if that meant winning and succeeding together. But if winning meant shutting down our company, or succeeding meant quitting and doing something else, that would be OK, too.

Michael’s brand of caring is tough, charismatic, and wise. And it’s unmistakable – he takes a ton of time and makes real sacrifices in the service of caring for his students.

I hope this was a fun read for other YC alumni, and hopefully brought back some good memories. It definitely did for me – my cofounder and I had a lot of fun reminiscing. If you’re considering applying to YC, do it – it will be a great experience and it will make you a better founder.

✌️

Invited To Interview At Y Combinator

2019-10-17T00:00:00+00:00

The Art in Res team got some pretty exciting news last night.

For those who don’t know, Y Combinator is a startup incubator-slash-accelerator that funded and advised some pretty awesome companies, including some household names like Airbnb, Stripe, reddit, and Dropbox.

They also run Hacker News, a digital haunt for the Art in Res team and startup & tech folks everywhere.

And, they’re also widely considered to be one of the best, if not the very best, of the incubators and accelerators. See this, this, this, this, this… you get the picture.

But there’s something much more exciting - and personal - about being invited to interview with y combinator partners in a couple of weeks (!) in Mountain View.

When my cofounder John Friel started his first tech company, xhbtr, it was Paul Graham’s essays that provided the early blueprint. And when we started Art in Res, the same was true for me and Noni.

PG’s essays are blueprints, pep talks, life lessons, mental models, reference points, maps and warning signs, all conveyed in bite-sized plain english. And the philosophy that runs through those essays (and is perpetuated by Y Combinator today) is deeply ingrained in Art in Res (we hope).

There’s a lot of stuff we could be doing better. But, for pulling an application together at the 11th hour at the recommendation of a friend and recent alumnus, it’s really cool to know we’ve already beat some long odds.

Here’s hoping that the interview process treats us kindly - but at least it feels like we’re on the right track. Wish us luck! (And if you are interested in helping us prep for the interview process, email me!)

Also, massive, gigantic thanks to

for all the help!