Jun 16 2026

Chatting with my DNA – a fascinating personal use case and a deep rabbit hole

This one has nothing to do with running a company. But it’s one of the most fascinating things I’ve done with AI, and I fell down such a deep rabbit hole that I have to share it.

My friend Chad Dickerson — former CEO of Etsy, now an executive coach — mentioned that he’d uploaded his raw DNA file to an AI and started asking it questions. My reaction was something like: wait, you can do that?

You can. Here’s how. If you’ve done 23andMe (or Ancestry.com — the process is similar), you can download your raw DNA data file. It’s just a text file full of SNPs — single nucleotide polymorphisms, the genetic markers that make you you. Upload that file to Claude, and voilà — you can chat with your DNA.

I cannot overstate how much of a rabbit hole this is. I’ve spent hours in there. Hours. And I keep going back. I’ve enhanced the single, now very long, chat by adding in lab results, my current supplement and perscriptions, and family history over time. It’s just so cool. And yes, everyone, I know Claude isn’t a doctor. More on that below.

Medical

I started with straightforward physical questions. Why do my eyes water when I use my phone in the dark? What are my genetic markers for a handful of conditions that run in my family?

Key learnings:

  • Genetics explain mysteries, but family history trumps individual SNPs. A single genetic marker rarely tells the full story. But when you layer it on top of what you know about your family’s health history, patterns emerge fast.
  • Protective genetics are real. You might carry markers for something concerning, but other genes can counteract or mitigate those risks. The picture is more nuanced than “you have the gene or you don’t.”
  • Context matters more than the gene. The AI was most useful when I gave it additional context — lab results, family history, current symptoms. The gene is a clue. The context is the case.

Wellness

From there I went deeper: sleep optimization, diet, exercise, stress management. Not generic wellness advice — advice grounded in my actual genetic profile. And the random: Why do I sleep terribly in Denver but fine everywhere else?

Key learnings:

  • Interventions need lead time. Some genetic predispositions mean you need to start addressing things earlier than the general population would.
  • The body’s signals need translation. Symptoms I’d been ignoring or dismissing made a lot more sense once I understood the genetic context behind them.
  • Sometimes one gene explains everything. The Denver sleep mystery? Solved completely by a single variant related to how my body processes oxygen at altitude. Years of terrible sleep on business trips to Colorado, explained in one answer.
  • “Healthy” habits are context-dependent. What’s good for the average person might not be optimal for your specific genetics. This isn’t just about family history — it’s about your genome.
  • Hierarchy of interventions matters. Not all changes are created equal. My DNA pointed to which interventions would have the highest leverage for me specifically.
  • Prevention works best before symptoms. Obvious in retrospect, but the genetic data makes it concrete and personal in a way that generic health advice never does.

The Bonus Round (Where It Got Wild)

This is where it went from interesting to head-exploding.

I asked my DNA to predict my leadership and management style. And I asked for career advice. I specifically noted in the prompt that I wanted the responses to come only from the DNA file and not from other information in Claude’s memory about me. I didn’t want it pattern-matching on “Matt Blumberg, CEO” — I wanted it to work purely from the raw genetic data.

The response on my leadership style was shocking. With one exception, it read like almost every 360 review I’ve ever received. The strengths, the blind spots, the tendencies under stress — all of it. From my DNA. I sent it to my executive coach and several long-time colleagues. All heads exploded. Every one of them said it was uncannily accurate.

The career advice was equally wild. The AI suggested seven jobs I’d excel at given my genetic profile. I’ve done four of them.

Let that sink in. Four out of seven career recommendations — derived purely from a text file of genetic markers — matched careers I’ve actually pursued over 30 years. No résumé. No LinkedIn profile. No context about my life. Just the code.

The Moral of the Story

In the endless debate between nature and nurture, your genes know an awful lot about you.

But here’s the important nuance: genetics doesn’t tell you what to do — it tells you what to pay attention to. The most valuable insights came from combining the genetic data with lab results, family history, and an honest assessment of my own behaviors and habits. The DNA file is the starting point, not the answer key. Again, Claude is not a doctor, and as my own doctor says, he likes to treat the person, not the numbers. But with all of this data in it, Claude knows more about the whole me than any doctor ever has or ever will, making it a powerful personal resource, a fun pasttime, and a useful aid to my actual doctor.

If you have a 23andMe or Ancestry account gathering dust, download your raw data and upload it to Claude. Give yourself an hour. You won’t stop at one.