Your First Idea Doesn't Matter

On Saturday I shared a deep dive on how positioning drives conversions.

Anyway, here’s today at a glance:

Opportunity → Surrogacy Marketplace
Framework → Customer Problems Stack Rank
Tool → Codeant AI
Trend → Prompt Tuning
Quote → Your First Idea Doesn’t Matter

Opportunity: Surrogacy Marketplace

What does it mean when you find a market that has a lot of red flags (see Rex’s points below)… but is clearly growing and will likely be able to support a venture scale business?

To me, that smells like opportunity. Doing the hard things that other people don’t want to can be a competitive advantage.

When I was building my last startup, we started a global community by running a coliving house in Tulum, I was literally going to the grocery store to pick up giant jugs of water for the house after we lost potable water due to flooding.

Alternatively, if investors won’t fund it because they see too many challenges, these are the types of markets that can create very lucrative cashflowing businesses.

Surrogacy may be one of these markets:

Read on X

Rex wrote more about it and some other healthcare trends in his newsletter, but what would a surrogacy marketplace or vertical SaaS startup look like?

🧠 Framework: Customer Problems Stack Rank

You know that saying “everyone is fighting a battle you know nothing about?”

Turns out it’s true for your customers, too.

You might have visibility into how your customers are using your product or service, but you probably don’t have an understanding of every problem they’re trying to solve across their entire business.

This framework, dubbed CPSR by Shreyas Doshi, helps you get that additional visibility.

It’s extremely simple:

Ask your customers to rank the problems they’re actively trying to solve from most important to least important.

Where does the problem your product or service is helping them solve fall on this list?

The lower it falls on their list, the harder it will be for you to convince them to spend more and the faster they may be to churn if things aren’t working. If you’re selling B2B you can take it one step further by asking to get the same list from team members at the company who aren’t your primary contact.

Shreyas shared this as a product management framework, but it’s arguably even more applicable to founders.

🛠 Tool: Codeant AI

At a startup you need to be scrappy in order to move fast and that often means cutting corners when it comes to code quality.

Don’t pretend like it’s not true, we all do it.

Codeant AI is a VS Code extension backed by Y Combinator that adds optimizations, documentation, and scalability to your code automatically. So you can move fast without breaking things.

I just discovered them this week and am a big fan.

📈 Trend: Prompt Tuning

You’ve probably heard about prompt engineering, but how about prompt tuning? It’s been exploding throughout this year:

But what’s the difference between the two?

Basically, prompt engineering is about crafting better inputs to get better outputs from an AI model, while prompt tuning involves actually adjusting the model itself to better suit certain tasks or types of input.

From a business perspective, this turns generalized models into infrastructure for everyone else and implies that open source models will have an advantage in the long run due to their customizability.

Aadit Sheth writes the huge prompt engineering newsletter, Prompts Daily, and we caught up yesterday about prompt tuning. I’m including his analysis exclusively for premium members:

Aadit mentioned a huge benefit of prompt tuning is that it enables businesses to customize their favorite AI models for specific tasks without extensive of expensive training.

For example let’s say you own an online book store (Amazon 1.0). With prompt tuning you could tweak a general AI model, rather than hiring scarce talent and developing your own from scratch for tens of millions of dollars, to provide personalized recommendations like how big companies are able to.

Over time this will continue to become more and more cost effective for even the smallest companies. In fact, we’re already seeing some evidence of a pricing war and race to the bottom play out.

Aadit sees opportunities across all verticals for disruptive companies that focus on tuning generalized models, he mentioned fashion but emphasized it’s not limited to that at all.

💬 Quote: Your First Idea Doesn’t Matter

Founders put a lot of pressure on themselves to get things right. This can be healthy, as pressure leads to great things and for many people helps them create their best work.

But don’t confuse progress with perfection.

Rather than being paralyzed by a fear to get it wrong, focus on learning as much as possible as quickly as possible.

Once you make this mindset shift you’ll open yourself up to failing (and failing fast) because learning comes from gathering feedback and iterating on your offer to users.

Ideally each experiment is more effective than the previous one, but not all will be and that’s ok.

This idea applies in particular to your first idea on a project: