Frequently Asked Questions
Find an answer
Search questions or choose a category, then expand an answer.
What is the decision layer?
It is the path by which a request gathers evidence, reaches the appropriate person or rule, and becomes authorised work. An approval, quotation or exception can involve several records and systems. HEPO models and executes configured steps so their state and results can be examined. Read more.
What does agentic mean here?
A configured process can combine tool operations, AI-assisted interpretation and human decision points. Its allowed actions and gates must be explicit. The process may execute authorised steps; a suggestion from an assistant does not itself supply authority for a consequential action. Read more.
Does the assistant scrape screens?
HEPO describes its assistant around registered, typed capabilities and process operations. For the workflow being evaluated, inspect the actual tool or interface used and the permissions it carries. This is the basis for explaining the action, rather than treating a visual impression of a screen as an API contract. Read more.
Can existing screens work with new processes?
How should we compare it with automation tools?
Use one real process: inspect how each approach obtains evidence, represents identity, handles human decisions, performs external effects and exposes failure. HEPO connects orchestration with its own operational records and registered capabilities. A demonstration is more useful than assuming every other tool works the same way. Read more.
Can we test a change before it goes live?
Use the supported sandbox or simulation path for the process being evaluated. Define what is simulated, what uses test data and whether any external effects are disabled. Inspect the state and results before enabling the live operation; a successful rehearsal is not a guarantee of production behaviour. Read more.
Where does process evidence come from?
Can an integration write back to an ERP?
Can processes use service accounts?
Background tasks and external integrations may use scoped execution credentials. Human decisions should retain their actor context, while the executing service and external identity remain distinguishable. Review permissions across each step rather than assuming the same credential is used everywhere. Read more.