Describe to automate.
How AI Colleagues get built – fast.

An AOP is to an AI Colleague what an SOP is to a human employee — a structured, versioned, executable playbook that describes exactly how a process should run. The difference: it's written in plain English and auto-bound to the right tools, APIs, and helper AOPs at authoring time.

Numbered steps, plain English, with tool calls inline using @ syntax. A typical Workday Time Off AOP runs 8 steps: fetch leave balance, confirm leave type, compliance check, dates and holiday check, mandatory comments, manager approval, apply leave, and escalate if blocked. Human-in-the-loop on approvals. Automatic fallback to a ticket when blocked.

Because variance is poison for SLAs, compliance, and audit. LLM improv looks great in demos and breaks in production — every run behaves differently. AOPs are explicit, deterministic rails: the process graph is written down. The LLM only fills in reasoning within those rails. Tools are deterministic too — each one knows exactly what system it hits and what fields it can change. No hallucinated API calls.

AOP Studio is the authoring environment inside Leena AI where business users create, test, and publish AOPs. It's how AI Colleagues actually get built — not a canvas, not a flowchart tool, an authoring environment built for procedures.

Business teams. HR builds HR's AI Colleagues. Finance builds Finance's. Procurement builds Procurement's. The people who own the process write the AOP. IT stays in the loop on governance, security, and policy — not in the critical path of every change.

No. You describe the process in plain English or upload an existing SOP — PDF, DOCX, flow diagram. The AOP Creator decomposes the steps, maps the tools, and writes the conditional logic. If you want to drop into the underlying AOP source to fine-tune, you can. Most people don't need to.

Days, not months. Authoring, testing, and publishing happen in a handful of sessions — not over multiple sprints. The Creator does the heavy lifting on step decomposition and tool mapping; you focus on getting the policy right.

Yes. Native connectors to 200+ enterprise apps, plus custom REST APIs, MCP servers, and A2A. AOPs sit on top of your stack — they don't ask you to replace it.

No rewrite tax. AOPs read like SOPs because that's what they are. Upload the document you already have. The Creator turns it into something executable without you reformatting anything.

Tools are the atomic capabilities an AI Colleague uses to act in the world — fetching a record, posting to Slack, opening a ticket, deactivating an account. Leena AI ships 1000+ pre-built tools across 200+ enterprise apps. The AOP Creator auto-binds the right tools as it parses your process. You can override, add custom ones, or call your own APIs.

Yes. AOPs can call MCP servers and orchestrate Agent-to-Agent (A2A) workflows alongside native tools and custom APIs. Open by design. Your stack, our orchestration.