I launched a product called Polymath earlier this week.

The core idea is that your LinkedIn and resume do not know you very well. But your AI does.

So Polymath will let you run analysis on what your AI tools (Claude, ChatGPT, Claude Code) know about you. It will run this analysis completely locally, and on your own Claude Code or Codex subscription. If Polymath thinks you're good, it will get you opportunities in the real world. Today this would be intros to smart people, and referrals into great companies.

This sounds like a recruiting agency with some bells and whistles. It is, right now.

However I have been building in the education and hiring space for 10 months now. I have pivoted through 4 different ideas before this. And I think this has a shot at changing education by navigating around the dysfunctional education institutions of the world instead of through them.

And I know this puts me at risk of sounding like every other delusional San Francisco founder you may have spoken to before, but I would love a chance to show you what the market taught me, and maybe you will see it the way I do too.

What the market taught me

In October last year I launched a product that would convert book PDFs into 1.5 hour long interactive audiobooks. I and a few of my friends used this and loved it. It completely replaced Kindle and Audible for me.

This was the first product I ever built that felt like early PMF to me, and a few of my friends agreed. All of us being avid readers.

But the market taught us why only a single digit percentage of people in the world read regularly. The reason is that it's not mission critical. This is true for both adults and students. People simply did not prioritise continuous learning much, it got pushed to the bottom of their TODO list and then out of it. This is why book summary companies like Shortform and Headway have tens of millions in ARR but not billions.

This taught me that a genius AI tutor would not get wild PMF, not unless it provably helped people get concrete outcomes, like acing a test or getting into Stanford.

Then I tried to go outcome first. I tried to hold 3 week long apprenticeships to teach high schoolers in India to build cool things with AI that would help their resumes stand out.

The response was lukewarm at best, I learnt how they're very busy preparing to apply for university and did not have the time to devote to this. There is an opportunity cost associated with breaking out of the current system, and that's part of why the system is so difficult to change.

I then tried to build something smaller, and more directly connected to outcomes. I built a 2 hour AI-native assessment, and tried to route smart students directly into startups. I learnt that a single 2 hour assessment with AI was not nearly enough signal, and the startups did not want to trust a third party assessment not personalised to them. They wanted domain-specific assessments and skills.

So to show companies these domain specific skills I offered a week long apprenticeship co-designed with some friends doing applied AI engineering at OpenAI. I got commitments from recruiting managers at Harvey, Sierra, Decagon and Cursor that they would look at these students' reports once they were in.

I tried to get college students to join it to learn and show their domain skills as an applied AI engineer. The best students told me they would never give a week of their time to something like this. This commitment to see their reports was not a good enough outcome.

This showed me a few important constraints that one has to build around:

1. If the upfront time commitment is long, the best people will not use your product.
2. If the outcomes are not very visceral, most people will not use your product.

So I decided to build Polymath.

Your AI knows you well already. You can approve and send that information in a trusted, controlled way to us immediately. And we will directly get you concrete outcomes, without you doing anything else.

But why could this change education?

It's such a complex space, many have tried and failed. All the VCs hate it. What makes this different?

Let me tell you what I think of education today.

All of education and hiring has reached a schelling point. A schelling point is an unspoken equilibrium that people get into. For example, if you ask 2 people to meet in New York City on 31st July, they would likely pick Times Square at noon. It's the most obvious way to implicitly coordinate without actually coordinating. The education space is in a schelling point.

Imagine you're trying to hire someone for a role. It is very difficult for you to measure every person's ability in your specific domain when they apply for the position. You interview five and work trial two. It's simply not scalable enough. So the university degree is the most obvious available option.

But this makes universities dysfunctional. It's no secret that the new grads are not very well equipped for the workforce.

I think the problem is the incentive structure of universities, and the way they make their revenue. The fees are collected upfront and for attendance. Their revenue is completely decoupled from the quality of education they give you, or the actual jobs you end up with.

Their revenue comes from their reputation. This is something that takes hundreds of years to build and decades to fade. This is an incredibly slow feedback loop. No wonder education has not changed.

It is a measurement problem. We cannot measure ability in domains effectively at scale, so we rely on loose proxies, and everyone is in debt and not at their full potential.

My bet is that if someone could evaluate exactly how good anyone would be at a given job, and employers started trusting it, the free markets could make education much more competent.

So that's the goal with Polymath. With zero upfront effort we add immediate value to exceptional people. Soon in the future, students can continue working with Polymath to learn domain skills and demonstrate their competence in those domains to get them jobs.

But it's a losing battle, one may say

AI will automate all of knowledge work and much of everything else in the next decade. You're building a recruiting agency. What then.

In this world the transfer of money will stop being a good proxy for the value of human actions, AIs will do almost everything better than us. Humanity will need to find a new way to assign value and meaning to human action.

These are thorny problems. I can't pretend I have the answers, but I do think we can make some directionally correct inferences.

It seems important to have an AI that sees you fully. An AI that is yours, that works with you to help you reach your maximum potential, that helps you find meaning. An AI that understands how capable you are and fights for your place in this world.

This is what the 4 pivots were all about. This is why everyone telling me education is uninvestable could be right, yet I don't mind giving years of my life to this and failing.

This future is important to build.

I'm doing this alone right now. If this resonates with you I would love to work with you. Contact me at akshay@polymathsociety.us.