This is Part 5 of a 7-part series examining Ribbit Capital's Token Revolution thesis. Previous installments covered the overview, identity (KYA), memory tokens, and stablecoins. Today we map how vertical token systems are replacing traditional outsourcing across every major industry.

The Quiet Collapse of Traditional Outsourcing

Something broke in 2025. Not loudly, not in headlines, but in the quarterly reports of the world's largest business process outsourcing companies.

BPO contract value dropped double digits globally. The Information Services Group, which tracks every major outsourcing deal, reported that new BPO contract signings fell 12% year-over-year. Accenture, Wipro, Infosys, and Cognizant all adjusted guidance downward. The Philippines, which built a $30 billion BPO industry employing 1.4 million people, began seeing its first sustained decline in contract volume since the industry's inception.

The surface explanation is "automation." But that word obscures what is actually happening. Automation has been promised for decades. This is different. This is token factories eating vertical industries alive.

What Ribbit Actually Said

The June 2025 Token Revolution letter devoted an entire section to what it called "vertical token systems." The argument was precise: industries that built infrastructure around human knowledge workers would be disrupted by companies that could tokenize that knowledge instead.

The key insight is the distinction between different types of tokens. Generic language models are trained on generic data. But vertical token systems capture domain-specific knowledge that took industries decades to accumulate. That knowledge is scarce. It is proprietary. And the companies that tokenize it first will own their verticals.

Ribbit offered a taxonomy of how this plays out:

Stage 1: Tokenization wedge. An AI company enters a vertical by solving one specific, painful problem. Abridge converts doctor-patient conversations into clinical notes. Harvey converts legal documents into structured analysis. Suki transcribes and codes medical procedures. Each wedge seems narrow but creates access to something invaluable: the real-world feedback that turns generic AI into expert AI.

Stage 2: Expert token accumulation. Through millions of interactions, corrections, and refinements, the company builds a token corpus that no competitor can replicate. Every time a physician corrects an Abridge note, every time a lawyer edits a Harvey output, the system gets smarter. The data becomes the moat.

Stage 3: System expansion. Once trusted for one task, the token factory expands into adjacent workflows. The note-taking company becomes the billing company becomes the prior authorization company becomes the operating system for clinical data. The legal research tool becomes the contract generator becomes the deal negotiator becomes the infrastructure for transactions.

This is not speculation. It is already happening.

The Seven Verticals Getting Eaten

We mapped every major industry where vertical token systems are displacing traditional BPO work. The pattern is consistent: AI agents enter through a single wedge, accumulate expert tokens, and expand until they own the workflow.

The combined market for these verticals exceeds $250 billion annually in outsourced services alone. Add internal operations and the number approaches $500 billion. That is the addressable disruption.

The Healthcare Case Study: Abridge as Template

Ribbit's letter uses healthcare as the clearest example of vertical token systems in action. The analysis is worth unpacking in detail because it reveals the playbook every other vertical will follow.

Medical documentation is a $40 billion global market. Physicians spend an estimated 2+ hours daily on documentation, contributing to burnout rates exceeding 50% in some specialties. The industry responded by creating an outsourcing infrastructure: medical scribes (human note-takers), offshore coding centers in India and the Philippines, and a complex ecosystem of documentation specialists.

Abridge entered this market with a single wedge: AI-powered clinical documentation. The physician speaks during patient encounters, and Abridge generates structured notes. Simple. But the magic is in what happens next.

What makes this model lethal to BPO is the economics. A human medical scribe in the Philippines costs $8-15 per hour. They handle maybe 10-15 encounters per day. The total cost per encounter runs $5-10. Abridge processes encounters for a fraction of that cost and improves with every interaction.

But cost is not the real weapon. Speed is. A human scribe delivers notes the next day. Abridge delivers them in minutes. For a healthcare system, that means physicians sign notes during the encounter rather than staying late or losing documentation entirely. The productivity gain justifies the entire deployment.

Ribbit predicts Abridge will not stop at documentation. The same token system that generates clinical notes can generate prior authorization requests, patient summaries, referral letters, and billing codes. Each expansion captures more of the workflow. Each workflow generates more expert tokens. The flywheel accelerates.

The Legal Disruption: Harvey and the Death of Document Review

Legal process outsourcing was a $20 billion industry built on a simple arbitrage: American lawyers bill $500+ per hour, Indian lawyers bill $30. Discovery and document review migrated offshore. Law firms booked the spread.

That arbitrage is collapsing. Not because Indian lawyers got expensive, but because AI got cheap.

Harvey, the legal AI company that raised $80 million from Sequoia at a reported $700+ million valuation, processes legal documents at machine scale. Contract analysis that required a team of associates for a week now happens in hours. Due diligence that cost six figures now costs thousands.

The numbers are stark. Thomson Reuters reported that its CoCounsel AI handles contract review 60% faster than human teams with higher accuracy. Allen & Overy deployed Harvey across its global practice and reported 25% efficiency gains in the first six months. Kirkland & Ellis, the world's highest-grossing law firm, built an internal AI system that processes millions of documents monthly.

The disruption follows the same pattern as healthcare. Harvey entered through document review. But the expert tokens it accumulates through every legal interaction - every correction by a partner, every nuance flagged by an associate - compound into something more valuable than any offshore team can match.

Now Harvey is expanding into contract drafting, legal research, and transaction support. The wedge becomes the platform. The platform becomes the operating system.

Financial Services: The Compliance Collapse

Banks globally spend over $40 billion annually on compliance. This includes KYC (Know Your Customer), AML (Anti-Money Laundering) screening, regulatory reporting, and audit preparation. Much of this work lives in offshore operations centers in India, Poland, and the Philippines.

Ribbit's portfolio companies are positioned across this entire stack. Plaid owns financial data access. Stripe owns payment identity. Persona handles identity verification. Unit enables embedded banking. Together, they represent the infrastructure for what the letter calls "compliance as a service."

The transformation goes beyond cost reduction. Traditional compliance is retrospective: collect documents, review them, file reports. Token-based compliance is continuous: every transaction carries its own verification, every customer interaction generates trust signals, every system state is auditable in real time.

This is what Ribbit means by "KYA will be bigger than KYC." When AI agents conduct financial transactions, they do not need to be verified once. They need to be verified continuously. The compliance infrastructure that handled annual reviews is being replaced by systems that verify every interaction.

The India Question: What Happens to 1.4 Million BPO Jobs?

No analysis of vertical token systems is complete without addressing the human cost. India's IT and BPO industry employs over 5 million people, with 1.4 million in business process services alone. The Philippines employs another 1.4 million in call centers and back-office operations. These are not abstract numbers. They represent the economic foundation of entire cities.

ISG's February 2026 report showed troubling signals. BPO contract value was down double digits in 2025. Renewals were happening at lower price points. Scope was shrinking as clients moved routine work to AI and retained only complex, judgment-intensive processes.

The optimistic narrative is that AI creates more work than it destroys. BPO providers pivot to "Intelligent Process Outsourcing" - managing AI systems rather than doing manual work. Human workers move up the value chain to handle exceptions, manage relationships, and provide the judgment that AI cannot.

The pessimistic narrative is that this transition happens too fast for labor markets to adapt. A 25-year-old in Bangalore or Manila who spent five years becoming an expert medical coder faces a job market that no longer values that expertise. The feedback loops that made them valuable are now captured by algorithms.

Ribbit does not take a position on this transition. The letter simply observes that it is happening and that the companies positioned to capture expert tokens will define the next decade of enterprise software.

The Real-World Asset Connection

Vertical token systems connect directly to the asset tokenization we examined in Day 4. BlackRock's BUIDL fund, now managing $2.2+ billion in tokenized treasuries, demonstrates how financial assets become machine-readable. But the same infrastructure applies to every industry vertical.

Healthcare records become health tokens that patients control and share selectively. Legal documents become smart contracts that execute automatically. Insurance policies become programmable agreements that adjust in real time. Supply chain documentation becomes a token trail that verifies provenance instantly.

This is the convergence Ribbit predicted. Vertical token systems generate expert knowledge. Asset tokenization creates programmable value. AI agents become the interface between them. The result is not just automation of existing processes. It is a fundamental restructuring of how industries operate.

What This Means for the Token Revolution Thesis

Day 5 of this series reveals why Ribbit sees the Token Revolution as more than a cryptocurrency phenomenon. The companies building vertical token systems are not blockchain companies. Abridge, Harvey, Tractable, and Suki are AI companies that happen to be building token-like assets: proprietary datasets that compound in value through feedback loops.

The blockchain connection comes later, when these expert tokens need to be verified, transferred, or monetized. A healthcare AI that accumulates clinical knowledge could tokenize that expertise and license it to other systems. A legal AI could create verifiable credentials for the quality of its analysis. An insurance AI could generate auditable proof of its underwriting decisions.

This is the infrastructure Ribbit is building. Not just AI companies in isolation, but an interconnected network of token factories that can transact with each other. The pieces are in place: Stripe for payments, Plaid for data, Persona for identity, Unit for banking, and now Harvey, Abridge, and their peers for vertical expertise.

What Comes Next

Tomorrow's installment examines the attention and truth deficit: how token systems address the crisis of information quality that threatens to undermine AI's promise. If anyone can generate convincing content, how do we know what to trust? Ribbit's answer involves a fundamental restructuring of how value flows through information networks.

The vertical token systems we mapped today are building the expert knowledge layer. Tomorrow we explore how that expertise gets verified and why attention itself is becoming the scarcest resource in the token economy.