Sequoia's Product-Market Fit Framework

Hey y’all — here’s today at a glance:

🔗 Houck’s Picks

My favorite finds of the week.

💡 Opportunity: Document Collector

AI agents are coming to help you do busywork, and one of the first use cases I’d love to see is document organization and collection.

Throughout the year, documents get scattered between your Downloads folder, desktop, inbox, Slack, etc etc etc etc

What if you could give an AI a goal, like Shahed does below, and have it go find and compile all related docs for you in seconds?

Ideally it would also label the docs and rename them in a coherent way for this specific task (it can always remember to rename them to something else once the task is complete, if you’d like).

This sounds simple but it would cut down literally billions of hours of busywork if adopted widely. You could do a simple MVP with individuals, but the real market is in B2B. You could eat a large chunk of Notion’s market (last valued at $10 billion) if it’s successful.

🧠 Framework: Arc Product-Market-Fit

Every founder wants to reach PMF, but what are the paths to get there?

Sequoia Capital’s Arc accelerator team put together this framework for early stage founders that outlines 3 ways to reach PMF:

1. Hair on Fire → You’re solving an obvious but important problem in, likely, a crowded market and do it better than anyone else (example: Zoom, during the pandemic).

2. Hard Fact → You challenge a conventionally accepted fact, and demonstrate an alternative (example: Superhuman).

3. Future Vision → You skate to where the puck is going. Your customers don’t even realize how much they need your solution until they use it (example: Tesla).

📈 Trend: Synthetic Data

When I was on Uber Eats’ data science team in 2018, we were able to quickly train new machine learning models by creating thousands or millions of extra datapoints.

This is synthetic data — it’s not how users are interacting with your product, or any other real dataset. Instead, you use those real sources to create more data that roughly follows the patterns found in the real data.

In 2018 it was a relatively niche process, even in academia, but it has explored as transformer-based models can handle increasing amounts of data.

There are some massive markets here. As demand for synthetic data increases, there will be opportunities to:

• Identify opportunities for companies to leverage synthetic data

• Create self-service tools for companies to create synthetic data

• Resell synthetic data

• And likely more that are still hard to grok

💬 Quote: Narrow Your Audience

If your aim is to build a massive company, but you may think you should go after a massive market.

But this is often a mistake.

Going broad means:

• More battles to fight

• More noise and less signal from users/customers

• More capital investment required

Not only do all of these make you less likely to be able to convince investors your startup is a smart bet, you’ll also just be more likely to fail.

This is one of the areas Paul Graham, Peter Thiel, and Nikita Bier (below) agree:

Start with a small market and dominate it.

Only after this, expand either horizontally or vertically.

You’ll have the momentum, cashflow, and bandwidth to do so more effectively.

💡 How I Can Help

🚀 Advertise in this newsletter to get in front of 80,000+ founders.