On deciding what to work on for the next 10 years

How I pivoted from an AI problem to another in 2022

On September 20, 2022, I attended a tech conference with around 50 attendees. During the conference, they held a breakout activity where any of the attendees could start round-table discussions on any off-topic subject. There were 6 round tables set up, each with 6–7 seats.

I started a round-table about Building a Second Brain¹. The whole thing lasted ~40 minutes. By the end of my session, there were over 30 people gathering around the table, with more standing than sitting, even organizers of other round-tables were joining.

I was there because my friend and I had received an invitation a few days earlier. At the time, we were working on a computer vision startup. By the end of this brief session, I knew two things. First, my vision-based idea hadn't received anywhere near this level of interest at the conference. Second, it seemed everyone was silently suffering from information fatigue in one way or another.

1. Pick a problem

Although at this point AI vision felt like a natural progression of my career, my pull towards starting a second-brain product was much higher. Heading out of this room, I knew it was over for the startup I was on at the time.

August 27, 2022, 7:30pm – First startup

I kept working on both in parallel for a few months until I eventually decided to stop working on the vision company and pursued the second-brain one. During this time, and through 2023, this choice has enabled me to talk with lots of people in the personal knowledge management (PKM) space, and get deeper insights about consumer software.

While I believe there are lots of ways to start a startup, and although I’m no where near a ‘startup advisor’, I can probably pinpoint three of them: the fast way, the worst way, and my experience.

The fast way

Freelance for 18-24 months. Working for similar clients, solving similar problems, developing a deep understanding of their workflows and pain points. Ultimately you’d find a step –or more– of the pipeline that could be productized, i.e turned from a service to a product, enabling you to build once and distribute twice.

The worst way

Brainstorm ideas. Which I get the reason why young folks with high energy prefer to do it: Lower barrier of entry, and the exciting thrill of jumping headfirst into building something with friends. However, searching for ideas is considered one of the worst ways to start working on a problem.

My experience

Solve a personal problem. This comes from noticing problems that constantly bother you for so long. My problem was ‘overthinking’ which was a direct cause of information fatigue. And I found that ‘knowledge management’ was not only a possible solution but also one of a few things that had been quietly simmering at the back of my head for a few years.

The thing about working on a startup around an interest of yours (solving a problem to you) is that you won’t bother the long hours. If working to solve the problem feels like play to you, you’ll outperform everyone who does it for work.

Also, in the early days, most success from working long hours happens in spikes. Meaning a handful of rare events scattered over a long timeline. These are events where you get ‘lucky’; and they don’t happen often at the start. So picking a nagging personal problem makes sure you wake up as equally pumped on a day that’s not special, where nothing ‘magical’ happens.

Luck is just ‘opportunity meeting preparation’. And since opportunity doesn’t have a predictable way of happening, the only way to predict luck is to simply be ready (years of prep). So by way of aligning years of work around a specific focus, luck tends to happen more naturally. This is what’s often called ‘serendipity’ – where compounding from relationships you build and specific knowledge you get, leads to more of them.

So solving the right problem, at the right time, with the right people could be a wormhole to a world of possibilities.

2. Make sure it’s the right problem

I believe that, in startups, a job is just another word for ‘solving a problem’. Getting paid –by employers, investors, or markets– is the equivalent of getting rent for real estate inside your head; as you get better at what you do, you’re raising your cost for rent.

The type of the real estate is the business domain. The size of the real estate is how invested you are in the problem. The more head real estate you allocate to a problem, the better solutions you’re going to have for it, and the higher you get paid in the process.

In that sense a ‘problem’ becomes the intersection of what the world needs, what you're deeply curious about, and what you're capable of solving.

So what about the ‘information fatigue’ problem?

At some point, I realized that:

• Smart people are just normal people who know how to effectively manage their time and information

• All knowledge is connected, and the way we do our best work is by effectively managing this knowledge

• Brains have limited power connecting and organizing knowledge, so we look for external tools and systems

A data visualization of every node in a second brain over a year, animated as it is being created. You can see connections between thoughts & information, and how they emulate neural connections in a real biological brain. A tool becomes an extension of the mind.

Given the overwhelming amount of information today, this is clearly a problem for many others. So searching for ways where software can help makes lots of sense. Now add to that how AI can help, there exists an infinite number of possibilities to move forward.

“Everyone should have the chance to gather information and ideas, turn them into valuable knowledge, and use that knowledge to improve their own lives and the lives of others. Our biological brains are fundamentally limited, and we need to harness technology to help us think faster, better, and more clearly.”

– Tiago Forte in The 10-year vision

3. Keep working on the problem

Startups are a decade-long game. Most dividends from building good products are paid in 7-10 years. Less than that if you’re lucky, and about 10 if you’re realistic.

So it's not just long hours of work but also long years of commitment. So by definition, what you’d want to be doing 10 years from now should align with the kind of work needed to continue working on the problem, because it’s unlikely that you’d stop doing it anytime soon.

As for deciding what you want to be doing years from now, and although trivial, I find lots of people talking about a very simple exercise: Write it on paper.

I found the idea of writing down my values, serves as a roadmap that guides my decisions. At the very least, it helps stay on track for when I get distracted.

In an essay called What to do with your life, author Julian Shapiro, writes that just by listing values you care about, you’re able to effectively filter what projects to work on.

I did this exercise.

The values I wrote down were wisdom, freedom, and community.

Based on these values, and in alignment with the problem I picked, it felt natural for me to write down ‘Build aesthetic software’ and ‘Write prolifically’ as the focus for the next decade.

At least as long as it's there, I got a reminder; a compass.

Am I heading the right direction?

When was the last time we asked?

“People fail to realize that career success isn't an end state. Success is simply having the freedom to focus on an ongoing grind you enjoy.”

– Julian Shapiro in What to do with your life

I’m starting a “How to build a second brain” series, where I discuss how to organize and make sense of digital information using modern tools. This post serves as an interlude. Next post is going to be about information capture. Hope to see you around.

---

Thanks for

Ved Shankar

,

Matt Cyr

,

Ahmed Abdelsalam

,

Tahsin Khan

, and

Celeste

for reading drafts of this.

---

Let me know in the comments what are some problems from your world that are worth solving?