A perfect storm is hitting finance: AI that passes the CFA, PhD-level analysis on your iPhone — yet buy-side firms are still using Excel like it's 2009.

My semi-retirement ended when I realized that buy-side firms need a translator between AI capabilities and their investment workflows. And  LaTour AI was born.

But first, let's rewind.

I was promoted to Managing Director at BlackRock at 31 years old. Then I did what no one does — walked away for a decade to raise my daughters.

That decade gave me the distance to see what others are too busy to notice.

In January 2025, I launched "Future-Proof Your Career with AI" — the must-read newsletter for ambitious professionals seeking pragmatic AI use cases.

I have a Computer Science degree from Yale. I spent 15 years analyzing quant funds. And the last decade as an Internet entrepreneur.

Turns out my weird resume is a feature, not a bug.

Now I help buy-side firms save 10+ hours a week using AI — from automated research to instant earnings analysis to LP letters that write themselves.

Visit www.latourai.com to book a consultation.

The era of 1 person billion dollar companies has arrived. 

Remember when Zuck swooped up Instagram for billions when the app reached escape velocity?

No one knows the terms of this “acqui-hire.”

But the pace is unreal. 

OpenClaw (née ClawdBot aka MoltBot) burst onto the scene 3 weeks ago. 

Since then it’s been downloaded over 200,000 times by AI noobs and enthusiasts alike. 

Few outside the coding community had heard about Peter Steinberger.

Now the viral AI assistant will likely become a core offering inside OpenAI’s portfolio. 

What does this mean for you? 

Well, for starters, it will probably get easier to install very soon.

But more importantly, it shows the exponential pace at which things are moving.

And if you think it's just for coders and hackers, I urge you to reconsider.

This speed of change is coming for knowledge workers across every single industry.

I write a weekly newsletter called Future Proof: Your Career with AI to help you stay up to date with all these changes.

You can sign up in the first comment.

Khemaridh (Khe) Hy

graphical user interface, text, application

If you think you're behind with AI, you're wrong.

Just make sure you're a red dot.

There are 8.1 billion people on earth. 

Only 0.3% pay for AI. 

And 82% of American businesses aren't using it for ANY function.

But open X or LinkedIn and you'd think every company is crushing their AI strategy.

You're not late. 

You're absurdly early.
6.8 billion people have never touched an AI tool. 

ChatGPT has 900 million weekly users but only 35 million actually pay. 

And the business side is even wilder - only 4% of companies have "mature" AI capabilities. 

78% of executives openly admit they can't keep up.

This is the internet circa 2004 all over again. 

Back then, helping a business "get online" didn't require a CS degree. 

It required showing up, understanding the tech better than your competitors and creating value.

Now the red dots are the ones truly pushing the boundaries. 

They're building agents (without knowing code) and automating their research, analyses and business development.

Want to know how to become a "red dot?" 

Join our AI Coaching program - The AI Accelerator.

Link in first comment.

Source: @noahepstein_/X

Buckle up. We've hit PowerPoint AGI.

Stop your formatting immediately. Do not press Control M.

This week two major developments dropped this week that could finally free us from the soul-crushing task of slide deck creation. 

Claude launched a native PowerPoint plug-in that works directly inside your presentation software. 

NotebookLM now lets you edit and export presentations to PowerPoint format. 

I tested both extensively. 

Claude's plug-in excels at working within your existing brand templates and modifying decks iteratively. 

NotebookLM creates better-looking designs from scratch and has superior synthesis capabilities when working from multiple sources. 

Here are my key takeaways: 

1. Stop living in chat windows: If you're still copy-pasting prompts, you're missing the real leverage. Native integrations and persistent tools create compound value over time. 

2. Design still matters: While both tools can generate functional slides, you'll still need to refine the aesthetics and ensure brand consistency. 

3. The export function on NotebookLM is huge: Being able to edit AI-generated presentations bridges the gap between useful and actually usable.

Will you miss PowerPoint? https://lnkd.in/djz5Yh8k

Open article: Have we reached PowerPoint AGI? by khemaridh.substack.com

Mac Minis are sold out across New York City. Folks are scooping them up to set up personal AI agents. I refuse to buy one.

Instead I found an old laptop with a busted monitor, wiped the hard drive, and bought a $60 monitor.

I created a fresh Gmail account with zero external API connections, and set up OpenClaw.

I'm insanely paranoid about security.

Minimal permissions. Read-only access where possible. No 3rd-party plugins. I'm legging in slowly and deliberately.

My first project: networking. 

The claw now reminds me who I haven't spoken to, surfaces upcoming birthdays, and is starting to recommend talking points based on things I've read.

It took about 15 hours to set up and debug. (Thankfully, learning AI is *my job*)

But even in the first week, I'm saving 5 hours a week on tasks that used to require manual tracking, memory, and context-switching.

My approach isn't to go all-in. It's to skate to where the puck is heading. 

To learn by doing. To experience what a proactive AI actually feels like when it starts to understand your work, your relationships, and your priorities.

And it's awesome. I don't think I can go back to using regular interfaces.

---

Stay up to date with my weekly newsletter Future-Proof your Career with AI - sent on Thursdays and Sundays.

Sign up in the first link.

Congrats, you mastered prompting. But that's such a 2025 skill. Let me explain: 

When you copy-paste into a chat window and re-introduce yourself every session, you fail to create any leverage.

Enter Claude Code. 

First, it's true that it was originally designed for software engineers. 

I'm not a coder. Chances are you aren't either.

But that's no longer true. Anyone can "code" with the world's most spoken language: English.

Claude Code is a completely different game.

It lives inside your file system, reads your data, and builds tools that keep running long after you close your laptop.

A great prompt gets you a good answer once.

Claude Code gets you a tool that generates great answers forever.

We're talking transcript analyzers that profile prospects after every Zoom call

Email agents that surface your highest-leverage replies

And a second brain that's always enhancing your internal research.

I built Unfair Advantage to take non-technical professionals from AI-fluent to dangerous in 7 days.

It's $249, lifetime access, zero coding required.

Open article: Unfair Advantage: Master Claude Code (for non-technical professionals) by khehy.com

Unfair Advantage: Master Claude Code (for non-technical professionals)

Claude Code — the unlock for more time with your family?

Half the dads at swim practice have their coding agents fired up.

That's a real tweet from Matt Yglesias, and it describes my life perfectly.

I've been coding roughly 4 hours a day for the past year.

I don't know how to code.

Claude Code runs long, uninterrupted sessions. You describe what you want, and it builds it. With your voice if you want.

I've used it to build a suite of tools for business development, prospecting, and customer research.

Manager dashboards, data analytics, CRM workflows — all things I would have hired contractors or bought software for.

The Max plan is $100/month. It's the best $100 I've ever spent.

But here's the part no one talks about.

Being able to do all of this from your phone means you can kick off an agent while you're at the park, at swim practice, or putting your kid to bed.

You let the agent work on the long-term stuff. And you focus on the things only you can do.

That's the real unlock — it's not about coding more. It's about being present more.

---

Become a Claude Code power user, even if you're non-technical. Sign up for Unfair Advantage and you'll learn this powerful tool in 7 days.

Https://https://lnkd.in/gXbnFEgC

How did you go bankrupt?
Two ways. Gradually, then all of a sudden.

Unless you were on a silent meditation retreat, you read the viral AI post "Something big its happening" by Matt Shumer. 

He argues that AI is an exponentially powerful technology that is coming for everything you consider sacred: your job, dignity, financial assets and your kids’ futures.

It’s the ultimate AI Rorschach Test. It will violently reinforce your existing world view on AI. If you’re bullish on AI, it’ll make you more bullish.

And if you think AI will destroy the world, well that destruction has already begun.

One of Shumer’s key arguments is that AI dominates at coding. He’s right and the data overwhelmingly supports this:

1. Spotify declared on an earnings call that its best developers haven’t written a single line of code since December

2. Both the Claude Code and Codex GPT coding agents were predominantly built by themselves

3. A significant percentage of new code on GitHub is AI-generated

Coding is an activity that lends itself particularly well to AI. The accuracy of its outputs is directly verifiable. 

Shumer’s bull scenario hinges on the belief that all of mankind’s problems (and desires) can be distilled into software problems. 

Healthcare, finance, education, poverty eradication, entertainment are all fixable with software.

But turning non-software problems into software problems is easier said than done.

This is not how much of the finance industry operates.

Between old ways of doing things, legacy tools (ahem, Excel) and a general apprehension to anything “coding-related” you can understand the hesitation (and difficulty) of redesigning an entire process to be more software-oriented.

The same could be applied to industries like healthcare, law and consulting. How can you turn your work processes into software problems?

I believe that Shumer is right that every single worker should be striving to push the limits on transforming their existing workflows into software problems that can be tackled by AI.

Yet I also believe tremendous human judgement, intuition and experience are required to integrate all of that information into an actual decision.

Read the full post in the first comment

Another day, another AI-driven market sell-off.

This viral post predicted a dark scenario for 2028.

But it starts with a rosier picture that: AI works *exactly* as it's supposed to.

It's so bullish, that it then becomes bearish.

Companies cut headcount, reinvest savings into AI, AI gets better, more cuts follow. 

Each company's decision is rational. 

The collective result is catastrophic.

They coined a term for it: "Ghost GDP" - output that shows up in national accounts but never circulates through the real economy. 

Because machines don't spend money on discretionary goods.

---
I'll be writing more on this theme later this week in my bi-weekly newsletter, Future-Proof your Career with AI (sent on Thursday and Sunday).

Sign up in the first link.

Everything you learned about AI in 2025 is already wrong. And we're still in February.

Last year, "using AI" meant picking a chatbot and getting good at prompting. That playbook is dead.

According to AI researcher Ethan Mollick, 2026 means understanding the three layers of AI: models, apps, and harnesses.

A few quick definitions:

Models are the brains (GPT-5.2, Claude Opus 4.6, Gemini 3 Pro). These are what the benchmarks measure.

Apps are the products you use to access the models (ChatGPT, Claude, etc.)

Harnesses are the systems that let AI actually DO things - use tools, browse the web, write and test code, manage files.

Here's why this matters: the same model behaves completely differently depending on its harness.

Claude Opus in a chat window is a very different experience than Claude Opus inside Claude Code, autonomously writing software for hours.

So here's what to do right now:

Stop defaulting to the basic model. 

Every platform hides the smartest model behind a menu. Manually select the most advanced option. 

Try at least one agentic tool. NotebookLM is free and a great starting point. Claude Cowork and Claude Code are the most powerful if you want AI that does work, not just discusses it.

Stop optimizing your prompts and view yourself as more of a delegator.

The gap between people who treat AI as a chatbot and people who treat it as a workforce is about to become very visible.

---
Every week I write a newsletter that helps you master AI so that you can future-proof your career.

It's called Future-Proof your Career with AI and you can sign up in the first comment below.

Stop using AI like it's 2025. I'm begging you.

Last year, AI was a ton of copy-and-pasting. That era is over.

In 2025 the big themes were model selection, prompt engineering and context scaffolding. You'd have long conversations with models and paste the outputs into your documents.

In 2026, we can kiss cmd-c, cmd-v goodbye.

The question is no longer "what model should I use?" It's "which product surface should I use?"

Chat is like Google Maps. You need turn-by-turn instructions. It's high-touch. It works, but it's a PITA.

The new tools are more like a Waymo. You give it your destination. You don't care what route it takes.

Claude Code and Codex write code. Cowork automates knowledge work. NotebookLM retrieves from your own sources. Excel and PowerPoint have product-specific plugins.

We're moving away from answers and towards outcomes.

https://lnkd.in/ggRhNMH6

Open article: Stop using AI like it's 2025 by khemaridh.substack.com

I paid my daughter $7 to learn Claude.

Yes, I broke one of the sacred rules of parenting.

Every parenting book tells you not to use cash to motivate intrinsic learning. But my 11-year-old wasn't interested in AI, and her school keeps telling her "AI is bad."

So I offered her $7 an hour.

Here's how it went.

First, I set her up with her own Claude login, the Pro plan, and the desktop app. I had her use WisprFlow for voice input and store her passwords in 1Password.

Then came model selection. I made it very clear there's a hierarchy and she should always start with the best model available.

Next was the AI safety talk. Not the birds and the bees — but don't share personal info, don't use it on homework without asking, and I need to be around while she's using it.

(Honestly, my biggest fear isn't job loss or privacy. It's that she'll lean on AI for emotional companionship.)

We started by having Claude "interview" her — her hobbies, classes, and interests. I explained that this builds what professionals call "context."

Her first project was building a game. One prompt. Claude built the artifact. Then I asked her to come up with 3 improvements on her own, and 3 more with Claude's help.

She even created her first GitHub account. A proud dad moment.

Her general vibe throughout? "How much more time until I get my $7?"

We burned through all her Opus credits in one session. I wonder if she comes back for round 2 this weekend.

Open article: I paid my daughter $7 to learn Claude by khemaridh.substack.com

The economy is K-shaped. The top 10% of earners now account for half of all consumer spending. You know what else is K-shaped? 

The AI Economy. 

Two groups, same economy, completely different realities.

People in SF are running multi-agent swarms, consulting Claude before every decision, building tools that would've been sci-fi two years ago.

Everywhere else, teams are still waiting for IT to approve Copilot.

I see the same K-shape in my own work.

The upper arm of the K is creative, adaptive, and thrives under uncertainty. 

They experiment. 

They're fine when things break. 

They don't wait for a playbook.

The lower arm needs structure, guidance from above, and freezes in the face of ambiguity. 

Not because they're less smart. 

Because they've been trained to wait for instructions.

The gap between these two groups is widening faster than anyone expected.

And here's the uncomfortable part: companies aren't sorting people into these groups. 

People are sorting themselves.

The question isn't whether your team is using AI. It's which arm of the K they're on.

Sign up in the first link.

From X: @kevinroose

---
Stay up to date with my weekly newsletter Future-Proof your Career with AI - sent on Thursdays and Sundays.

Block just laid off 40% of its workforce. Why? AI.

And the stock is up 23% in after-hours trading.

Here's CEO (and ex-Twitter founder) Jack Dorsey:

"Within the next year, I believe the majority of companies will reach the same conclusion and make similar structural changes."

Now, we don't know the full story from the outside. 

Block nearly doubled its headcount since 2020.

So is this really about AI, or is AI the convenient explanation? 

That's fair to ask.

But here's what I keep thinking about.

When ATMs were introduced, everyone assumed the same thing: bank tellers are done. 

Fewer tellers, fewer branches, fewer jobs.

The opposite happened.

ATMs made it cheaper to operate a branch. 

So banks opened more of them. And the tellers? 

They moved into higher-value services like financial planning and advisory. 

The number of branches went UP. 

The number of employees went UP.

The ATM didn't eliminate the bank teller. It changed what a bank teller did.

Now here's the question no one answered then, and we're not answering now: did ALL the bank tellers become financial advisors?

Some did. Some didn't. And that's the part we need to be honest about today.

AI will almost certainly create new roles that don't exist yet. 

But the transition isn't automatic. 

It requires training, personal agency, and a willingness to invest in people - not just tools.

---
Every week I write a newsletter that helps you master AI so that you can future-proof your career.

It's called Future-Proof your Career with AI and you can sign up in the first comment below.

Welcome to the SaaS apocalypse.  Software stocks just lost $1 trillion in market cap. AI is coming for their moats.

Here are my 5 Top AI Reads for the week:

1. The $1 Trillion SaaS Reckoning
A breakdown of the 10 software moats that LLMs are dismantling — and two rebuttals.

2. The Wonderful World of AI Plugins
MCP, Skills, Sub-Agents and Plug-ins are creating a whole ecosystem you don't need to be a programmer to use.

3. OpenClaw's Creator on the Lex Fridman Podcast
Three hours of technical, accessible and surprisingly philosophical conversation — one of the best AI pods this year.

4. Accenture Links Promotions to AI Log-ins
Big brother isn't coming — he's already tracking your AI usage (and reporting them to HR).

5. OpenAI Codex Launches Skills and Automations
The coding interface is evolving fast, and it's worth paying attention to this new form factor.

Open article: The $1 Trillion SaaS Reckoning by khemaridh.substack.com

taplio

Khemaridh Hy's Linkedin Analytics

Khemaridh (Khe) Hy

Performance ScoreKhemaridh (Khe) Hy's LinkedIn Performance

What is Khemaridh talking about?

Khemaridh (Khe) Hy's Best Posts (last 30 days)

Ready to grow your LinkedIn presence?

taplio

.css-6ytdur{background-image:linear-gradient(284deg, rgba(39, 43, 72, 1) 0%, rgba(33, 47, 183, 1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Khemaridh Hy's Linkedin Analytics

Khemaridh (Khe) Hy

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Khemaridh (Khe) Hy's LinkedIn Performance

What is Khemaridh talking about?

Khemaridh (Khe) Hy's Best Posts .css-1emoo84{font-size:var(--chakra-fontSizes-md);font-weight:400;color:var(--chakra-colors-gray-600);}(last 30 days)

Ready to grow your LinkedIn presence?

Khemaridh Hy's Linkedin Analytics

Performance ScoreKhemaridh (Khe) Hy's LinkedIn Performance

Khemaridh (Khe) Hy's Best Posts (last 30 days)