In partnership with

Hello to all ~10,000 of you from around the globe. Thank you for reading and sharing!

Nine months ago in Issue #267, I wrote about what I called the AI First Draft Dilemma. Clients arriving with AI-generated documents, false confidence, and a quiet erosion of the lawyer-client relationship. I said it would escalate.

It has.

This week two things landed in the same 24-hour window that, read together, tell you exactly where this is going and how fast.

The View From 30,000 Feet

Y Combinator is the startup accelerator behind Airbnb, Stripe, and Dropbox, and arguably the most reliable signal of where serious technology money is actually moving. They just published their Summer 2026 funding wishlist. Most people read it as a VC signal. I read it as a forecast.

One entry by Gustaf Alströmer, a YC partner, described a new category called AI-Native Service Companies. His framing is precise and worth sitting with. These are companies that don't sell software. They sell the service. Not a tool that helps people do their jobs. A replacement for the people doing the jobs.

The sectors he names: insurance brokerage, accounting, tax, audit, compliance, healthcare administration.

Now law isn't on the Summer list. But it was on YC's Winter 2026 Request for Startups. It explicitly read: start your own law firm, staff it with AI agents, and compete with existing law firms. That batch delivered. Several AI-native firms came directly out of it.

And compliance, which is on the Summer list, is already being done at institutional scale. Norm AI, which I’ve been covering lately, with Blackstone as its anchor investor and co-designer, has been running agentic compliance workflows for global banks, hedge funds, and asset managers including Blackstone and Citi for two years. The collective AUM of its existing compliance clients exceeds $30 trillion. The law firm is the next layer on infrastructure that already exists and already works.

The total spend on services is many times larger than the spend on software. YC knows this. That's why the money is moving here.

BTW: The line between a product, software, and services is blurring rapidly. This won’t matter so much for the buyers and consumers, but for those businesses building anything with AI, including firms, this blurring has massive business model considerations. It's harder to “bolt on” AI-native offerings to a legacy business.

The View From the Trenches

While YC maps the future, lawyers on X (Twitter) are living the present.

Ann Srivastava put it plainly this week. Every American lawyer in 2026 will receive this type of email. Client pastes a supplier agreement into ChatGPT. ChatGPT flags the indemnity clause, the missing termination right, the three-year lock-in. Client sends a three-page memo. Lawyer reads it three times. Forwards it to a partner with one line: "client lost his mind."

Except sometimes the client isn't wrong.

Srivastava's sharpest observation, and the one the profession least wants to hear: general counsel at Fortune 500 companies are running outside counsel's work through Claude and ChatGPT as a matter of routine. And sometimes the AI catches things the partner missed. A missing IP indemnity. A liability cap one-tenth of what the deal warrants. A forum selection clause that would land disputes in the wrong court entirely.

She doesn't say the GC fires the firm. She says the GC renegotiates the retainer. And starts wondering if there are better options.

I wrote about this pattern in my Human on the Hook series last month. The GC doesn't disappear as a client. They just arrive differently. Pre-scoped, with a preliminary risk assessment, and a clear sense of what they want done. The orientation work that used to generate the first 30 billable hours is gone. The on-ramp to the engagement keeps getting shorter.

M&A attorney Eli Albrecht described the other edge of the same sword this week. A seller ran a standard, favorable LOI through ChatGPT. ChatGPT flagged the non-compete as unacceptable, a clause so routine it appears in virtually every transaction of this type. The seller killed the deal. Albrecht's line on this is a great one to keep in mind: AI in its current form is like a person you meet at a dinner party who talks loudly and confidently, but upon scrutiny lacks nuance or strategic understanding.

Both things are simultaneously true. The AI catches real things real lawyers miss. The AI also tanks deals over phantom concerns and invents case law that doesn't exist. We all know the stories of lawyers getting sanctioned for citing cases ChatGPT fabricated. But it get much worse! See below, a claimant in Illinois filed 44 pro se documents trying to reopen a closed case, including citations to a case called Carr v. Gateway, Inc. that exists nowhere in legal history.

Srivastava's framing for all of this is helpful: the trust asymmetry hasn't disappeared. It has moved. Clients now understand 40 percent of what lawyers do, where they understood 5 percent before. Enough to ask uncomfortable questions. Not enough to know when the AI is hallucinating.

That is the most dangerous zone in any professional relationship.

What I Said in September Is Now Your Monday Morning

In Issue #267 I wrote that outside counsel were navigating a new kind of document review, one with hidden risks, awkward conversations, and new economic pressures. I gave you a checklist. I said the conversation with clients would feel awkward at first.

The awkward conversation is now weekly. The checklist isn't optional anymore.

Here's the call to action, stated as plainly as I can.

If you haven't run your last five client matters through Claude or ChatGPT and asked, "what would you challenge here," you are flying blind. Your clients are doing it whether you do it or not. The lawyers who thrive in this environment won't be the ones who dismiss the AI memo. They'll be the ones who already ran the same analysis, caught the same issues, and show up to the client conversation one step ahead.

The solo practitioner I wrote about recently won a client from two 15-attorney firms because he spent 45 minutes doing what they never bothered to do. That's the model. Preparation is the new competitive moat.

YC is funding the companies that will eventually replace the services that don't adapt. The Twitter thread is showing you what the clients of those services are already doing in the meantime.

The map and the trenches are pointing at the same place. The services being built right now will siphon off your clients and portions of your work, quietly, then all at once. The clients you already have are already auditing your work. The only move that makes sense in both directions is the same one: get ahead of it. Run your own work through the tools your clients are using. Build your own captive YC-desired business. Design process that establishes privilege before the prompting starts. Create the validation service for the AI-informed client who needs judgment, not orientation. The window to do this proactively is still open. It will not stay open.

 Talk soon again, Josh.

Someone just spent $236,000,000 on a painting. Here’s why it matters for your wallet.

Late last year, a Klimt sold for the highest price ever paid for modern art at auction.

An outlier sure, but it wasn't a fluke. U.S. auction sales grew 23.1% in 2025. The $1-5mm segment even grew 40.8% YoY.

When the S&P 500 finished its worst quarter since 2022 last month, diversifiers like bonds and bitcoin fell too.

Even with the turnaround in mid-April, analysts at Goldman Sachs and Vanguard have projected low-single-digit annualized returns from 2024-2034.

So, what kind of investment is largely indifferent to the forces driving everything else?

Masterworks lets you invest in shares of artworks featuring legends like Banksy, Basquiat, and Picasso.

  • Access to an asset class with historically low correlation and attractive appreciation (‘95-’25).*

  • Net annualized returns on 28 sold works held 12 months+ like 14.6%, 17.6%, and 17.8%.

  • $1.3 billion invested across over 500 artworks.

Shares can sell quickly, but my subscribers can skip the waitlist:

*According to Masterworks data. Investing involves risk. Past performance is not indicative of future returns. See important Reg A disclosures at masterworks.com/cd.

To read previous editions, click here.

Was this newsletter useful? Help me to improve!

With your feedback, I can improve the letter. Click on a link to vote:

Login or Subscribe to participate

Who is the author, Josh Kubicki?

Josh Kubicki teaches AI and the business of law at Indiana University Maurer School of Law and has trained over 3,000 lawyers on generative AI. He is the author of Brainyacts, read by nearly 10,000 legal professionals worldwide.

AI training, courses, and resources: kubicki.ai

Strategic advisory for firm leadership: joshkubicki.com

DISCLAIMER: None of this is legal advice. This newsletter is strictly educational and is not legal advice or a solicitation to buy or sell any assets or to make any legal decisions. Please be careful and do your own research.

Keep Reading