Workflow self-healing
When an RPA bot broke mid-run, the workflow stopped; AOPs pause, diagnose, and continue from step sixteen.
- Before
The RPA bot broke when the vendor moved a button. The team that built the workflow is half the size. The workflow stopped.
- What we did
Rebuilt it as an Agent Operating Procedure (AOP) with browser control. When a step fails, the run pauses, the agent diagnoses, and it continues. If infrastructure fails mid-run, the run resumes where it stopped.
- After
"The workflow was on step sixteen. The step failed. Here are all the logs." Then it kept going. Hard budgets cap the cost of any retry loop.
Heal-then-continue has limits. Some failures need a human. The run waits for one instead of dying.
Role automation, not task automation.
Rebuild brittle RPA as a heal-then-continue workflow
“He wrote the entire stuff. By itself. And I was there, and I was thinking, alright. If I had to write all this stuff, I mean, I would've spent days, days, days.”
“I send that to the programmer and he said, alright. I read your file. I copied the section of the code because he has the right interface, and it works. Wow. Just like magic.”
“Digital worker, it gets spun up, it gets given access to the toolkits that it needs. It gets given access to the data that it needs. Just exists for the period it needs to exist to perform the task. And then it gets torn back down again afterwards.”
“I like that direction, that concept of the agent is the agent. It becomes an entity in its own right with its own ownership and its own missioning, and you get rid of the gray area and the blurry lines.”
“A lot of people are telling us that, oh, I sold in this package because of this info... I secured this distribution because of the selling assistance help.”
“I am not a programmer. I cannot read a line of program in SQL or anything... The only thing that I'm doing is just writing prompts.”
“I finally got a chance to play around with the computer asset and build out the chatbot you mentioned. It was really impressive.”
“So we use the SDK, and we write that in Python, and it calls Athena majority by API. The results appear in the interface itself.”
“You guys are functioning better because the harness is tested for a longer period of time against enterprise customers.”
“It's like I'm running analysis that would have taken us, like, maybe a week, two. And thirty, forty minutes.”
Verbatim from customer calls. Customers anonymized.
Related products and stories
See this on your own work
Bring one recurring task, a sample input, and an example of work you would accept.