Video summary
The speaker utilizes advanced AI tools like Claude Code and various APIs to conduct deep retrospective analyses of his life, specifically focusing on two decades of angel investing and personal habits. By automating the processing of emails and historical data, he challenges his own narratives about missed opportunities or past successes against hard numbers rather than relying solely on memory. This process reveals a significant gap between perception and reality; while humans often believe they know their history accurately, AI analysis suggests that our self-perception can be off by roughly twenty percent due to the natural sanitization of stories over time. The technology allows him to test his personal "batting average" objectively, proving that many assumed failures were not missed opportunities at all and correcting misconceptions about past decisions regarding investments or daily choices like chocolate preferences.
Beyond simple fact-checking, applying AI to decision-making has provided crucial insights into holistic health by cross-referencing medications, supplements, and potential side effects with incredible speed. The speaker emphasizes that while LLMs can occasionally hallucinate, the ability to quickly verify information across multiple models minimizes risks of disaster in personal health management. In the realm of investing, AI serves as a powerful tool for running counterfactual scenarios—asking "what if" questions about past trades—to objectively scorecard thinking processes rather than just chasing returns. Although he acknowledges that most daily tasks are not worth doing manually anyway due to automation efficiencies, there is immense value in using these tools to analyze complex historical datasets that would take years of human effort to compile and review alone.
However, the integration of AI into his life also presents a psychological challenge regarding motivation and creativity. The speaker admits that seeing an AI generate high-quality content in seconds can drain his desire to spend thirty hours on similar tasks himself, creating a sense of obsolescence akin to top players feeling defeated by AlphaGo. Despite this, he views the technology as a superior co-pilot rather than a replacement because current models still struggle with generating original thesis ideas or novel concepts that genuinely engage an audience; they excel at refining and expanding existing thoughts but lack true creative genesis. He finds it fascinating that these machines never forget details from years ago if given permission to access conversation history, creating a "spooky" level of understanding about his preferences across various topics like writing, relationships, and nonprofit work.
Ultimately, the most transformative application he discovered was asking open-ended questions similar to those one would ask a close friend, such as requesting three to five rewarding career paths for the next five years. The AI's responses were not generic suggestions but included specific business ideas that aligned with his philosophical interests, prompting him to immediately pursue new ventures and rethink his future direction. This interaction demonstrated how an artificial intelligence could deeply inform major life chapters by synthesizing a vast amount of personal history into actionable advice. For listeners considering this approach, the key is to treat AI as a thinking partner for exploring creative professional avenues rather than outsourcing responsibility entirely, allowing the technology to help uncover paths that might otherwise remain hidden in one's own mental clutter.
Read the full video transcript
What are you doing now with the with the
eye stuff?
>> Well, I mean, I'm more curious to hear
your thoughts and predictions, frankly,
because I think you're better at it, but
I mean, I'm using
Claude code with Ferris APIs to do tons
of like inbox analysis and stuff.
>> Yeah, yeah, yeah.
>> Right? I'm doing a 20-year
retrospective analysis of angel
investing.
>> Mhm.
>> It's like, who made what introductions?
Which companies did I not reply to that
ended up being successes? Which did I
turn down that ended up being
really important.
>> really want to do that to yourself?
>> Well, I suspected it'd be worse than it
was. I actually have not I haven't
missed that many explicit opportunities.
I wanted to test my own stories against
data, right? Cuz I have all sorts of
stories
>> Right.
>> about why I did certain things and why
certain things worked out. And I have
certain stories about my
batting average. And I'm like,
but is it true?
>> Right.
>> Really?
>> Is it really true?
>> Let's look at some hard numbers.
>> The sad truth, and I have a buddy that
wears that the bracelet and you've seen
me with the necklace around that we
categorizes your AI and listens to you
24/7.
>> Yeah.
>> It's about 70-ish percent that we think
we know, but is actually what we know.
>> Yeah.
>> Out of the 100% of like what we think we
know, this is the truth. I said I wanted
a dark chocolate bar at 7:00 p.m.
>> Yeah.
>> It's like, no, you said it at 5:00 and
you said it this way.
>> And you said it was milk chocolate.
>> Yeah, exactly. So, it's always about 20%
off from where you actually think your
brain's at. Sure.
>> Which is brutal.
>> Well, plus, I mean, that's like last
week, right? If you're talking about 15
years ago.
>> I mean, if you listen to any genesis
story of any startup, you're like,
>> Oh my god.
>> Wait a minute now. Like, this is like a
startup comic working on material, but
he's been working on this one 5-minute
bit so long that now he believes that's
actually truth.
>> Yeah.
>> That I mean, the sanitizing and the
editing of these startup genesis stories
is hilarious, and there's no reason to
think that I would be or you would be
exempt from it.
>> Right.
>> Right, when you're telling your own
story, even if you're just telling it to
yourself.
>> So, what's the number one thing that
you've learned by applying AI to your
life in this fashion? Like, what's the
thing where you walked away and said
"Damn, that was insightful, and I'm
going to change my behavior, or I
learned something new about myself that
I wouldn't have if if I had not used
AI."
>> Well, I think from a holistic health
perspective, by holistic I mean
having enough data related to
medications, supplements,
predispositions, side effects, what
happened to me 2 weeks ago,
the LLMs have been incredibly helpful. I
mean, the picture that they get and the
speed with which they can deliver an
answer that I can interrogate is just
incredible.
>> What did you learn? Like, what was the
thing that you
>> I mean, honestly, it's mostly avoiding
disaster, right? It's like, are any of
these things contraindicated with one
another?
>> Could A, B, or C explain D?
>> And you have to keep in mind, these
things can still hallucinate, but
you can
I don't want to say eliminate that, but
minimize it by just fact-checking across
LLMs. There's that. I would say that
there's a lot of insight on, hopefully,
that that can translate to future
decision-making related to
investing, which investing for me is not
just amassing more chips. What's fun
about investing to me is it's a way to
scorecard your thinking and
decision-making.
It's just a very objective way
to decide if something was the right or
the wrong decision. And you can
fine-slice that, and there are ways that
you could maybe question that, but
if you're asking yourself, "Was I
thinking well last month?" That's not a
very helpful question. Where do you go
from there?
If you're logging maybe every decision
you make every day and then trying to
cross-reference outcomes with blah blah
blah. Like, yeah, but you're never going
to do that.
>> Mhm.
>> But when you're making relatively
frequent investments, you can do that.
You can also run counterfactuals, right?
What if I did the opposite? What if I
had not sold that? What if I had kept
that? What if I had done this? What if I
had done that?
>> Is that worth your time though? At the
end of the day, you could put everything
in the S&P 500 and just go to bed.
>> Well, there's that. I would say it's
worth it to me because I find it
interesting.
I actually enjoy the intellectual
exercise of it. But otherwise, I would
say with
in terms of like how AI has has impacted
me,
I would say that
the honest answer is not that much
because most [ __ ] isn't worth doing in
the first place. People are finding
very, very clever ways to expedite
automating workflows of all different
types and doing something well does not
make it important or worth doing in the
first place. So, there's a lot I think
the level of [ __ ] that is being done
just at a very fast, efficient rate is
skyrocketing. Yeah. But simultaneously,
there are definitely cases where
I look back at say this analysis of 20
years of stuff.
To do that manually would be impossible.
>> Sure.
>> All right, it would take me a year
full-time with multiple people to do
that. And with a Claude code Gmail API
and
leaving my computer running for a
handful of hours a few times, it's like,
what you get back is [ __ ] credible.
>> Yeah.
>> Like, it's unbelievable. And I haven't
even scratched the surface.
>> Yeah.
>> I will say also another way that AI has
maybe affected my life in a
net negative way, and I'm not We have
another mutual friend who maybe we
shouldn't name who feels very similarly
is
We'll leave that out, but yeah.
>> [laughter]
>> Jesus Christ. So, if you train
AIs on your writing,
they're really good.
>> Yeah.
>> And I think I feel This is a stretch of
a comparison, obviously, cuz I'm not
an adept like a world-class Go player,
but when AlphaGo defeated one of the top
Korean players,
>> Mhm.
>> he was kind of like, "I'm done. Like, I
don't find joy in this anymore."
>> Yeah.
>> If we're playing against machines.
>> Yeah.
>> And when I see these AIs very
beautifully, I'm not going to lie.
And they're We're like in the top of the
first inning, right? This stuff is going
to get so much better. Spit out stuff
that is so much better.
>> Yeah.
>> I mean, I can still write, but what they
can do in 30 seconds is what would take
me 30 hours. And I'm just like, "Fuck."
It really drains the motivation for me
to put in those 30 hours.
>> Yeah.
>> Why wouldn't it? Of course it would,
right?
>> Yeah, but in some sense, you can
consider it a really good co-pilot
because for it to come up with novel
ideas that would engage an audience,
that's still the holy grail where it's
not quite there yet, right? Like, it's
going to make you sound It's going to
button up your copy.
And it might expand upon it in ways that
you wouldn't. But it's not going to come
up with the original thesis for the
whole thing, right?
>> Yeah, it's going to have trouble with
the original thesis, but even there,
I think it does a pretty good job.
>> Really? It's getting better. I haven't
tried with writing stuff.
>> do a data dump and you're like,
"Create."
>> Oh, oh, you sent me that link.
>> Yeah, if you just do a data dump and
you're like, "Create."
>> Where you sent me that link, you were
like, "What should Tim do in the next 5
years?"
>> Oh, yeah, that was good.
>> really interesting. Tell people what you
did because like they might find this
They could apply this to their own life.
>> Sure. So, you could do this in whichever
model you're using, whether it's, you
know, Claude or ChatGPT or whatever.
If it knows you,
>> you can tie in your inbox, too.
>> Yeah, you can tie in your inbox. In my
case, I didn't do that, but if it has
enough history on you, you can just ask,
"What do you think I should do in the
next 5 years? What might be
some rewarding paths of exploration?" I
think I put something like that. "What
are three to five ideas
that you think could be rewarding career
exploration for me in the next X period
of time?"
>> So, for people listening, if they've
used AI for, let's call it, 3 to 6
months, and you've probably given it
several hundred things to think about,
yeah, it will span across those
conversations, as long as you turn this
on. It's I think it's on by default now,
but it used to be an opt-in thing, where
you have to say, "Allow the AI to look
cross-conversation." So, it has a
holistic understanding of who you are.
>> And the answers were [ __ ]
outstanding.
>> Yeah.
>> I mean, really, really good.
>> Yeah.
>> And I sent it to a few friends, sent to
you, I sent it to a few of my closest
friends, and they're like, "That's
pretty [ __ ] good."
>> Yeah.
>> I mean,
>> It was really cool. Some of the ideas I
was like, "Damn, you should do that,
dude."
>> Yeah.
>> It had this one business idea for you to
do, and it wasn't a book.
>> Yeah.
>> And I was like, "Dude, I texted you
back. I was like, that's awesome. Like,
go go build that."
>> Yeah. There were business ideas, there
were certainly kind of non-revenue,
but philosophically aligned ideas. It
was shocking to me. So, that's actually
a very good example of something that
has deeply informed what I'm mulling
over. As I imagine the future, I was
like, "Man, that actually is a really
good." Because, keeping in mind, I'm
asking questions about
things of interest, things I like,
things I don't like. I am asking
questions about
different scientific interests related
to Sci Sci Foundation, my nonprofit
foundation. I'm asking questions about
investing. I'm asking questions about
writing. I'm asking questions about
relationships. I'm asking questions
about organizing trips for friends. I'm
asking so many different questions. I
>> on the board of your nonprofit? I think
you're like secretary or something.
>> Yeah, something like that. [laughter]
>> I've never heard anything about that.
>> Yeah, I don't know. Maybe you were
honorably discharged. I don't know.
>> I did. [laughter] I never got any
paperwork around it. I've never heard
anything in like 3 years. I'm like,
okay.
>> Yeah, well, I said it was going to be a
light lift. It's a light lift.
>> It's a light lift.
>> [laughter]
[snorts]
>> That's a very good example, right? I
mean, that's that may be the best
example because if that even 10% informs
like a major next chapter.
>> Yeah.
>> 100%.
>> Like that's a big deal for me,
certainly.
And it makes me think a little bit about
podcast listeners, especially. Readers
also, but to a greater extent podcast
listeners who come up to me
and most most listeners I run into are
really great. And not I mean, there are
always a couple of weirdos, but most are
fantastic and they'll say something
often like, "I'm so sorry, you don't
know me at all and I feel like I know
you." And what I say a lot of the time
is actually, "If you listen to my
podcasts every week or even every month,
you do know me pretty well."
>> Yeah.
>> And then you think about
a machine that never forgets.
>> Yeah.
>> It's going to know you pretty damn well.
>> of course.
>> And it's spooky in a way.
>> Yeah.
>> But I started getting more out of the
LLMs
when I started asking questions and I
have to be I think a little careful with
outsourcing this
and absolving yourself of responsibility
to think about these things, but when
you ask it open-ended personal questions
in the way that you would ask a close
friend,
>> Mhm.
>> what do you think are three to five
creative ways I might explore things
professionally in the next 5 years?
>> Yeah.
>> As opposed to something that you think
is more suitable for a robot.
>> Yeah, yeah.
>> You get some really interesting
responses.
>> That's a great use case.