Two decades of automation, now supercharged by AI

Automation Anywhere has been orchestrating business workflows for over 20 years — long before the current AI wave. At HumanX 2026, WorkOS CEO Michael Grinich sat down with Chief Product Officer Peter White to talk about what changes when a company with two decades of automation expertise gets access to LLMs and AI agents.

A lot changes. But not in the ways most conference keynotes would have you believe.

From ocr to LLMs: what actually got better

Automation Anywhere works with banks, healthcare systems, and manufacturers to automate processes across cloud and on-premises environments — including decades-old legacy applications running on green-screen terminals. This is the unsexy, high-value work that keeps large organizations running.

When LLMs arrived, they expanded what was already possible. Document processing that once required specialized OCR and purpose-built machine learning models could suddenly handle complex extraction tasks — like parsing unstructured invoices or medical records with variable layouts — using generalized models. AI agents opened up entirely new categories of automation that were previously untouchable — not because the workflows were unknown, but because they were too unstructured for traditional rule-based approaches.

Customers are overwhelmed and want solutions, not technology

White shared a candid observation about enterprise buyers: they're exhausted by technology pitches.

"Agents are coming from every direction. I think a lot of customers are taking a step back and saying, 'I can't take another technology pitch. I need to know what solution you're going to provide that solves a business problem.'"

This is driving a real shift toward outcome-based selling. Instead of licensing a menu of technology components, Automation Anywhere is packaging pre-built solutions for specific workflows — like accounts payable automation — with pricing tied to business outcomes rather than platform access.

Enterprises aren't short on AI tools. They're short on AI tools that solve a named problem with a measurable result. The vendors who win the next phase won't be the ones with the most impressive demos — they'll be the ones who can say "this will reduce your accounts payable cycle by X days" and back it up with contractual commitments.

The reality check on AI hype

White offered a grounded perspective on the current AI discourse. While conference stages feature talk of fully autonomous AI departments, the reality inside most enterprises is more measured.

"A lot of organizations are still in the early phases. They're not automating entire departments away yet."

That gap between narrative and reality is worth paying attention to. The companies doing serious work with AI agents aren't replacing headcount wholesale. They're automating specific, well-understood workflows where the ROI is clear and the risk is manageable.

The anxiety is real, though. Employees worry about job displacement, and White highlighted Automation Anywhere's Pathfinder community program — focused on upskilling workers around AI and agents to keep them relevant as their organizations evolve. It's a pragmatic move: automation adoption stalls when the people closest to the workflows feel threatened by it.

Where AI is already delivering results in healthcare

The most compelling use cases White shared came from healthcare — and they're not theoretical.

Large hospital systems are using Automation Anywhere to accelerate oncology trial workflows. These organizations sit on massive volumes of clinical data, and the analysis work that previously took weeks can now be completed in a fraction of that time. Faster analysis means faster patient screening and enrollment, which means treatments reach patients sooner.

On the operational side, hospitals are using automation to speed up prior authorizations with insurance payers. This is one of the most painful bottlenecks in healthcare: a patient needs treatment, but the insurance approval process creates delays that can stretch for days or weeks. Automating the submission, status-checking, and follow-up steps of that workflow means patients get treated faster and hospitals recover revenue more reliably.

These aren't aspirational roadmap items. They're running in production, in environments where the stakes are measured in patient outcomes.

Where this is all heading

The AI agent conversation is moving fast, but the enterprises actually deploying agents at scale have a common thread: they started with a specific problem, not a technology bet. White's perspective — shaped by 20 years of watching automation cycles — is a useful corrective to the hype. The winners won't be the companies with the most agents. They'll be the ones whose agents solve problems people are willing to pay to have solved.

This interview was recorded at HumanX 2026 in San Francisco.