When agents replace human workers, companies need a unit of measurement for agent output. "Outcome Credits" emerge as the currency: one credit equals one resolved ticket, one generated document, one completed analysis. Salesforce's Agentforce conversation credits are the earliest example. This creates a need for credit infrastructure — metering, billing, settlement, and fraud detection for non-human work.
The companies building this infrastructure (billing platforms, metering APIs, settlement systems) occupy a position analogous to payment processors in the human economy. Stripe for agent work.
Every AI agent acting in an enterprise needs an identity: authentication credentials, permission scopes, audit trails, and revocation capabilities. This is a new identity management category distinct from human IAM (Identity and Access Management). An agent that accesses Salesforce, Jira, and Slack simultaneously needs cross-system identity that human SSO solutions don't provide.
The Cloudflare Zero Trust platform is one early solution. Specialized startups are emerging in "Machine Identity Management" — but the category is pre-revenue for most players.
The hardest problem: how does a CFO know if the $500K spent on AI agents last quarter actually delivered value? Human workers have performance reviews. Agents need equivalent measurement. This creates demand for "Agent Analytics" — platforms that track agent actions, measure outcome quality, calculate cost-per-resolution, and benchmark against human performance.
This is where the Agentic Accounting thesis connects: once you can measure agent value, you can properly classify the cost (R&D vs COGS) and justify the spend to investors.
The Digital Labor Economy is a 2027-2028 investment theme in its earliest formation stage. The infrastructure categories (billing, identity, auditing) are where the investable companies emerge first. Watch for acquisitions by Salesforce, ServiceNow, and Microsoft in this space.
Work-as-a-Service →Agentic Accounting →Cloudflare Toll Booth →