Deel is an HR and payroll platform, not a traditional tech company, yet it has delivered the kind of concrete, per-capita data that has been missing from the agent economics debate. By deploying its internal agent platform, Akai, the company claims to have added $140 million in ARR over 90 days without hiring a single new person. This shift, which saw revenue per employee jump from $130,000 to $215,000, offers a rare, measurable look at how automation impacts the bottom line, directly addressing the measurement problem our previous coverage identified.
This internal-to-external pipeline is becoming a clear industry trend. While tech giants like Cisco with its AI POD and Salesforce with AIforce have recently moved to commercialize their own internal tools, Deel’s transition is distinct. It highlights how a service-heavy business can successfully automate its back-office operations — handling over 250,000 cases per month — before packaging that capability for the broader enterprise market.
The numbers, as Deel reports them, are substantial. The platform currently manages over 250,000 cases per month and has saved more than 1 million hours across the company’s teams. Deel reports having over 10,000 live agents operating across finance, payments, and regulatory functions. The most striking figure, however, is the shift in revenue per employee — reportedly climbing from $130,000 to $215,000, a roughly 65% increase. For operations and finance leaders, this is a rare, tangible benchmark that can be compared against their own headcount economics.
The architecture behind these numbers relies on a split approach. The platform captures network traffic from recorded workflows to build connectors, prioritizing API-first integrations and using browser automation only as a last resort. For tasks requiring precision — calculating tax rates, verifying account numbers, matching payment references — the system uses deterministic rules rather than model judgment. AI reasoning is reserved strictly for genuine judgment calls, and any high-stakes decisions are routed to human operators via Slack or email. Given that Deel handles payroll and compliance across 150-plus countries, this accuracy architecture matters more than it would for a marketing automation tool.
It is important to note that all financial and operational metrics cited here are Deel-sourced and self-reported. There is no independent third-party audit available to verify the $140 million ARR figure or the specific revenue-per-employee gains. The company’s platform is third-party verified for security and management standards — including ISO 42001 for AI management systems, ISO 27001 for information security, and ISO 27701 for privacy information management — but the performance metrics remain internal claims. We are reporting these figures as they stand, acknowledging that they represent the kind of concrete data the industry has been searching for, even if they lack external validation. The Akai site was updated in September 2026 with refreshed metrics and customer testimonials.
Akai, which became available externally in May 2026, operates as a standalone product with its own contract, infrastructure, and pricing — enterprises do not need a Deel account to deploy it. The use cases are broad, ranging from payment reconciliation and FinCrime screening to customs documentation and government portal filings. By moving these processes to an agent-based model, Deel claims to have automated the equivalent of roughly 600 full-time employees’ worth of work.
For enterprise decision-makers, the question is no longer just about whether agents can perform tasks, but how they alter the fundamental math of a business. Deel’s experience suggests that the most significant impact of agent deployment may not be the total hours saved, but the ability to scale revenue without a linear increase in headcount. Whether these results are replicable in other vertical enterprises — and whether the self-reported numbers survive independent scrutiny — remains the central question for those evaluating their own agent deployments.