Process Automation on AIOP gives you two ways to ship: use the platform's pre-built modules and integrations, or extend it with your own services on the SDK. Either way, you get the full scaffold — ready AI model interfaces, identity, policy and audit baked in — and agents that can spin up projects, wire up tools and run processes through a chat interface, each with their own persona and scoped access.
Use the modules and integrations AIOP ships with — projects, tasks, agentic workflows ready to go — or extend the platform with your own services and integrations on the SDK. Both run under the same identity, policy and audit guarantees, with no parallel contract.
Ready interfaces to every major model, embedded into your processes — with the platform's identity, policy and audit guarantees in front of every call. The same scaffold is reused across every process that touches AI; you don't wire model access twice.
Ask an agent to spin up a project that evaluates balance sheets, sketch a small chat interface, wire up the tools it needs. The agent does the work — your job is to scope what it's allowed to touch and approve the result.
Every agent and assistant has its own persona, an explicit toolset and role-based access to data, steps and integrations. What an agent sees and what it can do is your decision — defined per role, enforced at the platform.
Agents triage inbound tickets, draft responses, propose remediations. The human handler sees the draft and the policy decision behind it — one click to send, edit or escalate.
Agents gather the context, prepare the approval package and route it to the correct signer chain. Every approval is signed and recorded; replay shows the full case history.
Scheduled agents run defined controls, file the results, escalate anomalies. Evidence accrues automatically — no parallel spreadsheet, no second source of truth.
Chat interface, the agent picks the tools, the persona, the role assignments — you scope what it's allowed to touch. Out-of-the-box modules become composable building blocks; custom SDK extensions plug in under the same contract.
Process Automation is the platform's Agentic Automation layer at work — a Managed Orchestrator for predictable, deterministic workflows and an Open Orchestrator for the ones you build yourself. Both inherit identity, policy, knowledge and audit from the layers below.
Process Automation doesn't live alone — it sits on top of the platform and reaches into every other layer that matters.
The eight-layer architecture underneath — the Agentic Automation layer is where process automation lives.
/solutions/platformWhat agents ground on — the same gated, cited knowledge layer your teams use.
/solutions/knowledge-managementThe role-based access that decides what an agent or assistant can touch.
/solutions/identityA common capability inside automated processes — extract first, route everything downstream.
/solutions/document-understandingBring one workflow — out-of-the-box modules or something we'd need to build for you on the SDK. We'll show you what running it through AIOP looks like.