The Right Idea Doesn't Exist

 Opportunity: Personalized Arxiv Paper Digest

One of the worst things about academic research is that so much of it never gets read. Even though it’s easy to argue much academic research is impractical for the real world, there are tons of genuine insights that never get widely distributed.

This happens for, mainly, two relatively obvious reasons:

• There’s no good way to discover new papers relevant to your interests

• The papers are often written in excessively dense language

LLMs would be great at solving both of these in a personalized way:

To build the system you’d need to ingest the abstracts of each paper on arxiv, and likely some other metadata (authors, at least). Run this process once per day so your data is always up to date, and then send a weekly digest to users with personalized recommendations and AI summaries.

Researchers would be a perfect early market for this, as they already want to stay up to date on findings in their chosen field.

An alternate initial GTM to expose this to a more broader audience would be offering this as a GPT in OpenAI’s GPT Store. The founder of Tweet Hunter and Taplio has seen some success this way for one of his new projects.

The lack of widespread readership of academic research has led to a lack of intellectual rigor over the years, the controversy with Harvard’s Claudine Gay is evidence enough of this, and something like this could help raise the bar back to where it should be.

🧠 Framework: Adjacent User Theory

Consumer social is hot again right now and I expect it to continue to be for the next 1-2 years, largely due to advances in AI (a16z put together a good early market map), so it seemed like a good time to talk about Bangaly Kaba’s adjacent user theory.

When Bangaly joined Instagram the company had 400 million users. The problem? That number was growing linearly, not exponentially.

He and his team identified a massive cohort of potential users who were low hanging fruit, adjacent users:

Among other reasons, Bangaly says teams tend to not focus on adjacent users because they’re busy focusing on power users.

This is an interesting perspective, traditional startup advice encourages founders to relentlessly focus on their most fervent supporters.

From my perspective, as long as you don’t have a leaky bucket with your power users then investing some time, thought, and resources into adjacent users can be viewed as another growth channel (as long as your product experience can support them too).

Just don’t spread yourself too thin, and if you have to choose then focus on your power users.

📈 Trend: Quitting Your Job

Remember when you had a job? Maybe you still do. If you do, you’ve probably thought about quitting.

Turns out basically everyone thinks about this at some point.

My read on this is that it spiked near the beginning of the pandemic due to more remote-friendly opportunities becoming available, and it’s stayed high as some companies have pushed employees to return to the office, even among rising layoffs.

Some interesting ideas in the space:

• An AI-tool that reads all digital media (social media, press, website, etc.) from a company and lets someone know if they’d be a good culture fit there (based on a survey they fill out)

• A hyper-personalized daily digest email of available jobs, curated by AI

• A Chrome extension that identifies gaps between someone’s current skills and what’s listed on a job page, and recommends online courses

• A SaaS tool for companies to identify likely quitters so they can begin collecting applications in case they need to hire

• A performance management tool for companies that tracks employee performance across multiple jobs during their career (with sensitive info redacted)

💬 Quote: Finding the Right Idea

Question: What’s the difference between someone who is waiting for a good idea, and someone who has too many ideas to be able to build them all?

Answer: They started building one of them.

You learn by doing, talking to people about their problems, forming hypotheses about how to solve their problems, and presenting those hypotheses back to people in the form of MVPs (and then iterating based on their feedback).

It’s impossible to have a grand vision that’s logical and based in reality if you aren’t hyper-engaged with that reality.

Once you start doing this, you’ll see problems everywhere, daydream about potential solutions, hypotheses to test, and creatively think of ways those solutions could end up becoming massive.

The interesting part is that your initial vision is almost always wrong. Markets evolve, preferences change, and once you’re on the journey you need to adapt to respond to those changes.

🔗 Houck’s Picks