A Letter to My Daughter (and Every Consultant Under 40)

My daughter is a consultant at a boutique firm. I'm a tech founder who builds AI products. I've been trying to find the right way to say this to her for months. This is my attempt.

Hey pal,

Keeping up with AI right now is like flying at Mach 3 at treetop level — you're moving so fast that the branches are hitting you in the face before you even see the trees. Every week, there's a new model, a new capability, a new industry getting restructured. Most people have stopped trying to keep up. I haven't, because I can't afford to. And honestly, neither can you.

I know the last thing you want is your dad sending you another AI video with a "you need to watch this" attached. But I've been trying to find the right way to say this for months, so here it is.

Two things happened recently that I can't stop thinking about.

A guy built a 200-line text file — not even a real program, just a set of instructions — that lets an AI agent review legal contracts. Not summarize them. Review them. The kind of work that used to require paralegals, junior associates, and a $15,000/year Westlaw subscription. Within days, $285 billion in market value evaporated from companies like Thomson Reuters and LegalZoom. Not because they became worthless overnight, but because the market suddenly realized their pricing model — charging per person, per seat, per hour — doesn't survive a world where one AI agent does the work of ten people.

The second thing: KPMG forced its own auditor to cut fees by 14%. Not because the audit was bad. Not because they found someone cheaper. Because AI exists, and KPMG looked across the table and said, "We both know these tasks can be done faster now. Your old prices aren't justified anymore."

That second one is what keeps me up at night. For you.

Here's what I'm not saying: AI is going to replace consultants. That's the lazy version of this story, and it's wrong.

Here's what I am saying: the way consulting gets bought and sold is about to change in ways that will compress margins, eliminate roles, and reward a completely different skill set than the one most firms are training for.

Right now, consulting runs on a pyramid. Lots of people billing hours. The client pays for bodies and time, and the firm makes money on the spread between what it pays you and what it bills for you. That model has worked for decades because the work genuinely required human hours — research, analysis, deck-building, implementation, project management.

AI doesn't eliminate the need for that work. It eliminates the need for that many people to do it. And the moment clients internalize that — the moment a CFO looks at a consulting invoice and thinks "you're using AI, so why am I still paying like you aren't?" — the negotiation changes permanently.

That's not theoretical. KPMG already did it. To their own auditor.

I've watched a lot of smart people talk about this recently. Peter Diamandis brought together a group that included Salim Ismail, Dave Blundin, and Alexander Wissner-Gross for a two-hour conversation about where it was all headed. One thing Wissner-Gross said stopped me cold: he invoked Moravec's paradox — the idea that the tasks we consider high-status and cognitively demanding are actually easier for AI to automate than manual, unstructured physical work. A machine can analyze a contract or optimize a supply chain model faster than it can load a dishwasher.

The implication is uncomfortable: the consultant's job may be more automatable than the electrician's. Not because the consultant is less smart, but because the work is more structured and data-rich — which is exactly what AI is good at.

They also talked about model release cycles collapsing. OpenAI went from releasing major models every 97 days to every 29 days. Anthropic's cycle is around 70 days. And the internal models — the ones companies haven't released yet — are months ahead of what you and I can access today. Three months of internal lead time in 2026 is a generational gap in capability.

The window where individuals and small teams can access near-frontier AI and build with it is open right now. It will not stay open forever. The best models will eventually go behind walls — proprietary, restricted, available only to the companies that built them. The next year or two is the window.

So What Do You Actually Do With This?

I've been thinking about that a lot, and here's what I'd tell you if we were sitting across from each other at dinner.

Stop using AI to polish things. Start using it to build things.

Right now, most consultants use AI the way the first car owners used cars — they bolted them onto horse-drawn carriages. You're using ChatGPT to clean up an email or summarize a document. That's fine. It's also the equivalent of a legacy software company adding a chatbot to their existing product and calling it "AI-powered." It's a bolt-on, not a transformation.

The people who will thrive are the ones who redesign entire engagements around AI. What does a project look like if AI agents handle the first 80% of research, analysis, and document creation, and humans focus on framing, judgment, and client navigation? The person who can design that workflow — not just use a tool inside the old workflow — becomes the most valuable person in the room.

Become the translator.

There's something called the "articulation problem." A VP tells you, "I need a better way to track the pipeline." That sentence contains maybe 5% of the information needed to actually build the right system. Good consultants already solve this — you interview, you map processes, you iterate until fuzzy business language becomes something concrete.

In an AI-heavy world, that skill becomes the scarcest one. Because the AI can build almost anything — but it can't figure out what to build from a vague request. The person who can sit with a client, absorb the messy reality of their business, and translate it into structured workflows, prompts, and guardrails that an AI agent can execute? That person has a job forever.

Get your hands dirty with the tools.

Not chatbots. Agents. There are tools right now that can generate production-grade code, operate across internal data sources, and chain multiple AI models together to solve problems no single model can handle alone. I know this because I'm building products on top of these systems every day.

You don't need to become an engineer. But you need a concrete, lived sense of what's possible so that when you're sitting with a client, you can propose something real instead of repeating talking points from a partner's slide deck. Build something small. Break it. Build it again. That experience changes how you think about every engagement.

Push toward outcomes, not hours.

This is the big one, and it's as much about your firm's future as yours. As fee pressure increases — and it will — clients will prefer firms that say "we'll own this KPI and design an AI-driven delivery model that hits it" over firms that say "we'll staff five people for six months." If you can gravitate toward engagements where impact is measurable and AI is baked into the operating model, you become central to the kind of work that survives the fee renegotiation wave.

The firms that figure this out will thrive. The ones that keep selling smart people plus slides plus billable hours — without deeply rethinking delivery around AI — will face margin compression and eventually consolidation.

I want to be clear about something. I'm not writing this because I think you're behind. I'm writing this because you're in a position to be ahead, and the window for "ahead" to matter is shrinking.

The consulting industry isn't dying. But the version of it that's been running since McKinsey invented the modern model is being restructured in real time. The data, the domain expertise, the client relationships, the accountability — those things still have enormous value. But the delivery model wrapped around them is what's breaking.

You have two edges that most people your age don't: you understand client-facing work at a level most engineers never will, and you have a dad who won't shut up about AI and will help you build anything you want to experiment with. Use both.

I love you. I'm proud of you. And I'm asking you to take this seriously — not because the sky is falling, but because the ground is shifting, and the people who move first get to choose where they stand.

Dad