Workflows combine Skills (reusable know-how from the Library) and AOPs (Agent Operating Procedures): versioned, scheduled, governed runs that turn your written SOPs into automated, auditable work.
Air-gapped in production at a global professional services firm for over a year.
A twelve-step SOP in a Word document. One person knows it. When they are out, it does not happen. When the vendor moves a button, the RPA bot breaks and the team that built it is half the size it was.
Automation tools solved the easy path and broke on the exceptions. Chat tools answer questions but do not show up on Thursday.
The gap is not intelligence. It is attendance.
Describe the procedure. Paste the SOP. Name the tools the agent may use, the sources it reads, the format it returns. Add template variables for anything that changes per run.
Pick the trigger. A schedule. An email address the AOP owns. A cell in a spreadsheet (=AOP). A file dropped in SharePoint. A message in Teams or Slack. A webhook. A web page that changed.
Set the gates. Mark the steps that need a human. Cap the run in dollars or model calls. Choose where approvals are sent.
Review the run, not the work. Each run writes its own summary. Approved states become restorable versions. Stale or duplicate procedures are flagged on the owner dashboard.
Workflows (Skills + AOPs). Agent Operating Procedures: versioned, triggered by a schedule, an email, or a meeting ending. Branch by entity or region. Run one per row with =AOP().
| Capability | People can | Agents can | Together |
|---|---|---|---|
| Workflows (Skills + AOPs) | Write the procedure, approve marked steps, roll back a version. | Run it on every trigger and every row; propose improvements. | Approval gates inside the flow; every run has a name and a version. |
Anything a person can do here, an agent can do with the same permissions, and both land in the same audit trail.
A shared sheet on SharePoint is watched at 7 AM. New rows are classified and formatted. On Friday the exec deck appears in three views: all, completed, over $1M. Roughly an hour a week back per person on the procurement team at a global manufacturer.
A shared inbox receives six weekly Excel files from partners. The AOP picks them up, runs the logistics team's existing script, and fills the 100-row output template. Nobody rewrote the Python. Nobody opens the inbox on Monday.
Paste a brief or email a folder to the AOP address. Every cite verified; overruled or superseded authority flagged; the review document returns with only the exceptions for a human.
Paste a brief or email a folder. Every cite verified; overruled or superseded authority flagged; the review document returns with only the exceptions for a human.
Software cost capitalization under ASC 350-40: one AOP classifies pull requests, a second escalates edge cases to a reviewer, a third computes journal entries. Thousands of pull requests in the first pass.
=AOP("contract-review", "file", A2) and drag down. 500 rows, 500 governed runs, each attributable to the person who dragged. Set the budget first.
“I'm very excited. This is something that I have been piecemealing with the limited tools that analysts get... rather than me at nine PM getting a failure email and deciding I'll just do it myself.”
“It's good to understand the capability of the platform. That how far, and how quickly you guys can get these things going [...] that has been impressive overall.”
“Lets us get a lot more out of our data than [...] We can sit on top of all of the data in [our data warehouse] as opposed to just, like, our little ingested, you know, three gigabytes semantic models. I think we have a lot of power here.”
“Gave Athena the spreadsheet, explained which column I was trying to figure out, and then ten secs it told me... exactly how the calculation was based on the other data.”
“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.”
“One workspace. All different, let's say, avatars of their output data. Right? Suddenly, it's dashboard. Excel is here. PowerPoint is when it's all together. It's very impressive.”
“Use the Athena platform probably closer to the concept and how it was designed, like, with the spaces architecture where you've got a set of primitives and capabilities, and you bring together the right set of capabilities to solve a task.”
“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.”
Verbatim from customer calls. Customers anonymized.
Budgets. Cap any run at N model calls or $X. Atomic across concurrent sub-agents. Fails closed if the ledger is unreachable.
Identity. Each AOP runs as a specific user or service identity for data isolation.
Versions. Approved states are restorable. Export as a package; redeploy across workspaces from Git, no click-ops.
Observability. Every step logged. Cost per AOP, agent, and user on the owner dashboard.
The first run exposes every ambiguity in your SOP. Expect to tighten the procedure a couple of times before it runs unattended.
A procedure the agent follows the same way every time, including steps, checks, and approvals, written in plain language.
Anything recurring with a known standard of done, such as reports, reconciliations, intake, and monitoring.
Yes. Approval steps pause the workflow until a named person signs off.
The workflow pauses, diagnoses the problem, and asks for help if needed.
Records every change by a person or an agent, rolls back one contributor's edits without losing anyone else's, and keeps the model, instructions, and sources behind each agent action.
How the platform is deployed: Athena's managed cloud, your cloud on AWS, GCP, or Azure, on-prem, air-gapped, or GovCloud. Same platform in every option.
Build, Work, and Data on top; 150+ connectors and every surface in and out; one map of the whole platform.
See ready-made workflows you can run on a schedule.
We will turn it into a workflow with you, agree the approval points, and hand it over.