RPA has failed.

Almost 20 years ago, the previous generation of RPA vendors set out to deliver a "fully automated enterprise". As UiPath's Daniel Dines proclaimed in 2016, "RPA would free humans to focus only on the things that truly matter."

They failed.

There are a few reasons for this failure:

1. The old approach to RPA was fundamentally unscalable. Companies found themselves creating hundreds or thousands of individual automation scripts, each requiring dedicated maintenance and oversight.
2. These automations were built on brittle, screen-scraping technology that would break whenever a target website updated its interface. Teams often spent more time fixing broken automations than they saved through automation in the first place.
3. Implementation required specialized consultants and lengthy professional services engagements, making true automation accessible only to the largest enterprises with the deepest pockets.

But most of all, the technology simply wasn't there yet.

This is changing.

New advances in AI in the form of Large Language Models (LLMs) allow us to move away from the RPA of the previous generation, to a new framework — Agent Driven Automation (ADA).

Instead of using hardcoded rules, we design agents that understand the high level objectives of the workflow they are automating, but can reason and decide how best to automate it as they run.

ADA puts humans at the helm of automation, not at its mercy. We envision a future where agents work alongside people, augmenting their capabilities while operating within clear guardrails and maintaining human oversight at every step.

The era of brittle, maintenance-heavy automation is over. We're entering a world where machines adapt to how people work, not the other way around.