When people hear what I do, they often ask about crypto, which stocks to buy, or where the S&P 500 is headed. I don’t know (and beware those who say they do). Almost no one asks about paperwork, how much to save, or how much risk to take—the unsexy parts. But I believe those matter more.
Here’s what I see as essential:
1) Keep estate docs current.
Will, power of attorney, health care proxy. Do you have them? When did you last update them? If you have kids, who would take care of them? When would they inherit? Dying without a will leaves a mess for whoever you leave behind. Don’t forget beneficiary designations—they supersede your will. Update them after major life events (deaths, divorce, etc.). Talk to an estate attorney about how to do this well.
2) Save enough for retirement.
What’s enough? That depends. For example, did you start working at 22 or in your 30s after completing a PhD? How many years do you want to work? And remember that careers can be cut short for unexpected reasons. A rough rule of thumb: save at least 10% of gross income each year. Most people need 15-20%. And you may need even more if you started late or want to retire early. That’s a lot, I know, but that’s what it takes. That level of savings also helps keep your spending in check—since you pay yourself first.
3) Insure your future earnings (human capital).
If you are in your 20s, 30s, or even 40s, you are probably your biggest asset. You have years of earnings ahead of you, but that could be disrupted by death or disability. Would your family have enough money? Are there others that depend on your income? Protect against early death and long-term disability with life insurance (low-cost term is best for most people) and a well-structured standalone disability policy.
4) Protect your assets.
You hit a cyclist with your car; a guest falls and hits their head. Bad things happen. Material assets? Max out your auto and home liability limits, and add a personal umbrella policy on top for further protection.
5) Take (some) equity risk.
Cash sitting in a bank account may feel safe, but its value is slowly eroded by inflation. Yes, stocks go down—sometimes 30%, 40%, even more than 50%. But over decades, a diversified equity portfolio has been the most reliable way to grow wealth after inflation. There are no perfect portfolios; only reasonable ones. The key is to have an allocation you can stay the course with—through the inevitable downturns.
6) But don’t blow yourself up.
Some people gamble with leverage and concentration and come out okay. Investing on margin, betting big on a hot sector, or putting a large chunk of money into a single company (especially the one you work for). But gamble and lose, and you can end up broke at age 50—that’s tough to recover from.
Many of us want predictions and hacks. When what we need is a plan, paperwork, protection, and patience. Unsexy and often complicated—but it’s what matters.
Two siblings just inherited $100 million. Real money.
They reached out to a recruiter, and he asked me to talk to them before taking them on as a client. The siblings wanted to build a family office to invest in startups and private equity. I encouraged them to slow down, think about what they actually wanted. Maybe they didn’t even need a family office. They didn’t like that answer.
Just sold a business or inherited a real chunk of money? Don’t feel pressure to do anything right away. (Taxes are the one thing that may have an actual deadline—talk to a good CPA early on.)
1) Put the money somewhere safe.
Short-term Treasuries or a money market fund. If you inherited an existing portfolio and it isn’t in cash, don’t sell it right away. That can wait until you come up with a plan. You aren’t in a hurry.
2) Ignore everyone telling you what to do.
The advisors and attorneys who read your press release and want to “help.” The CPA who refers you to their wealth team. Your high school friend, Bill, who you haven’t heard from in years but who just messaged you on LinkedIn with a special real estate deal. Even your sharp business school classmate.
Especially anyone saying you have to act now.
3) Think about what’s important to you.
What do you actually want? To build something new? To play tennis every day? To never worry about money again? This might be the first time you’ve had a real conversation with your family about money. What does this mean for your kids and grandkids? Who are you going to say no to?
If you don’t have a clear sense of what matters, other people will fill in the blanks for you.
4) Do your own homework.
Learn enough so no one can take advantage of you. Read a book or two (start with How to Think About Money by Jonathan Clements). Write down your questions. Interview a few qualified advisors—people who you identify and qualify. Take notes. Ask for written follow-ups if anything isn’t clear.
You might still decide to manage your own portfolio, but you’ll learn a few things along the way.
5) Come up with a written plan.
One you understand and can stick with. Not just for 1-2 years but for the next generation. Figure out what help you’ll need—tax, estate, and so on. Even if you manage the investment side yourself.
The dollar swings will be bigger now. A routine market drop can mean losing seven figures in a day—that hits differently. You won’t know how it feels until you live through it.
You might be tempted to skip ahead and just start investing. That’s what those siblings wanted to do. I hope everything worked out okay. I don’t know if they built their family office or not.
I do know that keeping wealth takes a different approach than building it. The hardest part is fighting the urge to do something right away. The money will be there when you’re ready.
Stages of Investing Enlightenment
Stage 1: picking stocks. How hard can it be?
Stage 2: mutual funds. I must need a professional.
Stage 3: index funds. This is working but feels too easy.
Stage 4: smart beta, factor tilts, tactical allocation. Who wants dumb beta?
Stage 5: private equity, hedge funds, tax-aware long/short. This is what billionaires do, right?
Stage 6: target date mutual fund. Ahh, it was this simple the whole time.
(I’m stuck between stages 4 and 5. Maybe I’ll get there someday…)
I played a lot of poker in college.
I wasn’t very good. But I still won.
This was 2003-04. Chris Moneymaker had just won the World Series of Poker Main Event after qualifying online. He turned a $40 entry fee into a WSOP seat and $2.5 million. Online poker exploded.
I knew the basics: starting hands, probabilities, position. That gave me an edge. Not because I was a great player, but because the competition wasn’t very good. Many of the new players knew even less than me.
I also chose my games wisely.
I stuck to low-stakes games—$25 and $50 buy-ins. Tables full of newer players.
The game got harder as bad players exited—a year later I was making less money despite knowing a lot more. It was easy for me to walk away from poker. I knew I wasn’t going to be an online poker player forever; it was mostly a way to pay for my last year of college. I cashed out and moved on with my life.
Even though I realized this early in poker, I didn’t apply this lesson to my career for years. Career choices compound. Every year you stay, walking away gets harder—more experience, more relationships, probably more money. You can’t just log off, even if you realize you are playing the wrong game.
I loved investing in private companies, but the game has gotten a lot harder since 2006 (when I started). It's now much more of a chasing game than an evaluating game. When I started, the job was finding good businesses and figuring out what they were worth. Now there's so much capital that the job is mostly trying to win deals (typically in very competitive auctions)—and praying the price you paid works out.
And each job changes as you get more senior—in private equity, from analysis to managing the deal process. Then convincing an investment committee to invest. Then raising money and running a firm. Each step changes the game. By the time you “make it,” it's a different job than the one you signed up for.
It's easier in poker to find a new table. But even there, most players don't spend enough time on game selection. It's much harder in life. We only get a few chances to select the games we play professionally. For some reason, I still keep choosing hard games. But now I do try to at least stop and think:
1) What's the game I'm actually playing?
2) Is it the same game I signed up for?
3) How good are the other players?
Because, although you don’t have to be great to win, you do have to be better.
Some professional advice fades. Some sticks with you forever.
When I was a first-year associate in private equity 20 years ago, a founding partner told me that most PE firms don’t reply to deals. They receive a deal, conclude it isn’t a fit, and never follow up.
How could that be, I wondered then. Deals are absolutely critical to the PE business. Many PE firms—then and now—tout their “proprietary” sourcing capabilities. They invest real time building relationships, searching for off-the-run deals, cold calling on companies (all things I’ve done myself)—and they don’t take a few minutes to reply when someone sends them a deal?
I didn't get it then. But he was right. It’s a small thing, takes just a few minutes, but many people don’t do it (not just in private equity).
He insisted that we always reply. A quick no. A sentence or two of feedback. Even better: “this isn’t a fit, but here’s what we are looking for.” He said it was an easy path to be top quartile in relationships.
We've all been on the other side when the stakes are higher.
The potential investor who stops replying after five meetings. The final round interview where you never hear back. The client who goes silent after you spent weeks drafting a plan.
Inboxes are even more crowded today. And it’s easier to ghost. The bar is low. Be the one who follows up.
(My exception: form emails. Life’s too short to respond to those.)
I think people underestimate how much the past has changed.
We see a new technology and “know” it is going to change the world. And it very well may. But we forget how much the world has already been transformed by one major breakthrough after another.
The sun sets and we flick a switch. We get a cut and forget about it by dinner. Electricity and antibiotics feel ordinary—just part of how the world now works. But neither existed 200 years ago.
We only know the after; we’ve forgotten the before. That skews our perception of how much the world has changed, leading us to overestimate how big the next big thing will actually be (on average, at least).
We also overweight the changes we’ve lived through, and forget the real breakthroughs.
What had a bigger impact? Email or the telegraph?
I remember when email went mainstream. You could send messages fast and cheap. Yet, before the telegraph, information traveled at the pace of a ship, a horse, or maybe a pigeon—miles per day, not per second. Messages took months to cross the country. The telegraph meant information traveled at the speed of light. Breaking news and market prices could be delivered within hours, not weeks.
Everything since—telephone, fax, email, text—has made sending messages faster and cheaper. But the fundamental breakthrough was the first one: the telegraph. It enabled things that were impossible before: synchronized financial markets and businesses that could span continents.
Maybe AI is a fundamental breakthrough (I think it probably is). But if we don’t really appreciate past transformations, what makes us think we can predict the shape of the next one?
@TheZvi Update: may have confirmed the mechanism. Web app reasoning_effort is set to 85/100 vs 99/100 in CLI. Plus, the thinking time "router."
https://t.co/ptL58bWiTP
@anthropic@alexalbert__ Did Claude throttle the web app with Opus 4.6? Claude tells me its reasoning effort is set to 85/100 (vs. 99/100 in CLI). They also decide thinking time based on perceived prompt complexity -- "simple" question = less thinking time. This maybe explains why it sometimes makes basic mistakes.
@TheZvi Should distinguish between Claude Code and Claude Web App (Chat). Chat seems to assume many hard questions are actually simple and doesn't think as much.
@anthropic@alexalbert__ Did Claude throttle the web app with Opus 4.6? Claude tells me its reasoning effort is set to 85/100 (vs. 99/100 in CLI). They also decide thinking time based on perceived prompt complexity -- "simple" question = less thinking time. This maybe explains why it sometimes makes basic mistakes.
Dear @anthropicAi, I would happily pay $200/month for a reasoning depth slider in the web app (same as the CLI). I like Opus 4.6 but, when it doesn't think, it misses or mixes up details that are right in the same chat. Simple questions can still require real thinking.
@emollick Specifically seeing two things: mixing up when things happened, and losing details from earlier in the same chat. The model can do it (4.5 extended handled this close to perfectly), the 4.6 router just doesn't know it should.
It is easy to know when markets are expensive but hard, even for the best investors, to make money on that. Ray Dalio, Peter Lynch, and Howard Marks all called the dotcom bubble. They were right—it was a bubble. When it burst, the NASDAQ collapsed 78% (the S&P 500 would nearly get cut in half).
The problem: their earliest calls were in 1995-96, but the market wouldn’t peak for another 4-5 years. The NASDAQ quadrupled (that’s 4x for those counting) from late 1996 to March 2000. That’s a lifetime in the investing business (especially if managing other people’s money). If you sold in 1997, 1998, or 1999, you could have made money on the roundtrip—but only if you were able to get back near the lows in 2002. Very few can pull off that immaculate timing; it requires both a well-timed exit and rebuy.
Soros lost $700m shorting tech stocks.
Julian Robertson closed his fund the same month the market peaked.
They were right. But early.
Even worse, countless investors sold early and then, as the market soared year after year, capitulated in 1998 or 1999. Druckenmiller, who ran Soros’s flagship fund, bet against tech in 1999. Then he reversed course and rode the market up, selling everything in January 2000. Even though he knew better, he couldn’t help himself and bought back into the market in March 2000, which turned out to be the peak. Almost no one was able to stay on the sidelines and watch everyone else get rich.
The challenge with valuation is that it tells us essentially nothing about returns over the next year or even three. Markets can be expensive and get even more expensive (as was the case in the mid to late 1990s). Or be cheap and get even cheaper. Valuation (choose your fighter) only works over 10, 15, or 20 years. And even then, it's a rough indicator of long-run returns, not a precise forecast.
That's why the best strategy, for most, is to have a reasonable allocation and stay the course. Perfect timing can work—that’s why it is seductive. But very few, even the best, can actually pull it off.
Are you better than an AI?
Many people assume yes. I think that's increasingly wrong. The AI models are already as good as—or better than—the average expert in many fields. Your clients may not know it yet. But I’ve seen firsthand it’s true.
To be better than the current AIs, you have to put in the time, do the thinking, be responsive—in other words, you have to care. Expertise (specialized knowledge) is no longer enough.
If you are an expert and care, I think you—as of today—can outperform the latest AI models. Expert but don’t care? It’s not even close. If you care but aren’t an expert, the AI is probably still better.
Working on a big regulatory filing last year, my lawyer had all the right credentials. Maybe it was the fixed fee or competing priorities but his deliverables were sloppy and inconsistent. ChatGPT and Claude fixed many mistakes and helped me largely rewrite the key filings. He may have been an expert but he didn’t care about my project. And the AI models ran circles around him.
My home computer was failing. After trying to diagnose the problem with AI, I reached out to technical support. They spent three hours on the phone with me to get everything back up and running. The IT person was an expert and also cared—he was better than the AI models.
From law to medicine, the AIs are already as good as the average expert. And you don’t have to worry about whether they’ll reply to your email, care about your project, or surprise you with a big bill. The AIs are always available, never distracted, and just about free.
And they’re getting better fast.
@alexalbert__ Feature request: web app projects need an open/closed toggle. Web app's main advantage over Code/Cowork is global context, but Projects break that. They can't see each other, main chat can't see inside them. "Help me prioritize my week" fails if Claude can't see across projects. ChatGPT already has this.
My dad read "significant blockage" on his scan results. He called. Pushed for an appointment. Finally, 3½ months later, the cardiologist saw him. And told him—exact words—"you're a ticking time bomb." I spent the last year navigating the healthcare system with him.
My lessons:
Old joke: The factory of the future has a man and a dog. Man feeds the dog. Dog keeps man from touching the equipment.
White-collar job of the future: Man feeds the dog. Dog keeps man from the computer. Computer does the work.
@cblatts Claude for Excel has changed how I error check/build spreadsheets. Worth checking out as well. This stuff is moving fast. Hard to stay on top of it.