Legacy Titans, Meet Your Meteor: AI’s $20 T Shot Across the Bow of Professional Services
This feels more like a fault-line map than a step function
Venrock partner Ethan Batraski’s new piece, “The Great Legacy Extinction: AI’s $20 T Takeover of Professional Services”, is worth a read.
It’s a brisk, data-rich look at why audit, tax, legal, P&C insurance, wealth management and the rest of the $20 trillion professional-services stack are about to be gutted and rebuilt by AI-native challengers. I’ve pulled the high-voltage points below, but you’ll want to read the original article for the full charge.
TL;DR in Five Bullets
$20 T in revenue, $0 in real tech advantage. The incumbents built moats on brand and billable hours, not on software. That leaves them flat-footed against firms whose cost of expertise trends toward zero as models scale.
The winning playbook is clear. Batraski spells out five levers:
Target rule-based, knowledge-intensive verticals
Train deep, domain-specific models
Design human-AI workflows that keep relationship capital front-and-center
Build compounding institutional memory
Ship transparent, auditable reasoning
Economics flip from “hours × rate” to “outcomes × software margins.” Think 100 %-plus gross margins and 5-10× client capacity per professional. That alone bankrupts the old pyramid model.
Three-phase change curve: augmentation → cannibalization → extinction. Incumbents get a short grace period while they bolt copilots onto legacy workflows. After that, the AI-first boutiques start strip-mining the mid-market, and finally the behemoths implode or reinvent.
Signals are already on the board. Deloitte and EY’s new agentic-AI platforms show the Big Four know what’s coming — but they’re still shipping feature-packs while startups are shipping business models.
Why This Matters (and to Whom)
Founders – If you’ve been looking for the next fintech-sized wedge, stop. “LegalGPT” or “TaxAI” isn’t a feature; it’s a firm. Pick a vertical, own the workflows end-to-end, and price on delivered value, not seat-licenses.
Incumbent partners – Your brand equity is melting faster than you can bill. Either refactor around AI-centric service lines now or prepare for a managed-decline strategy.
Investors – Treat these plays like vertical SaaS in disguise: fast CAC payback, sticky data moats, and escape-velocity margins. Early entries will compound harder than the first cloud-native ERP vendors.
My Take
We’ve seen this film before: mainframe → client-server, on-prem → cloud →... Each time, the disruptors didn’t just automate the old workflow — they rewrote the contract between producer and consumer. The difference now is that AI collapses both labor cost and cognitive bottlenecks simultaneously. That’s why Batraski’s extinction metaphor isn’t hype; it’s basic evolutionary math.
The power move is to treat expertise itself as software: capture it, compress it, and let it compound. Do that, and you don’t compete on day-rates; you compete on precision, speed, and an ever-learning knowledge graph that no human org chart can match.