Workflow platforms now let anyone put a language model in the middle of a business process, which makes it easy to build an automation that works on the demo input and fails quietly on the hundredth. This course teaches the design discipline first: triggers, actions, branching and data mapping that are common to every platform, and a clear rule for which decisions belong to deterministic steps and which to a model. It then works through four platforms on their own terms: Zapier AI steps, Zapier Agents and Zapier MCP; Make scenarios, Make AI Agents and the Make MCP server; the n8n AI Agent node with tools and memory, self-hosting in queue mode and the Sustainable Use License, and retrieval inside a workflow; and Microsoft Power Automate with AI Builder prompts, Copilot Studio agent flows and desktop flows for legacy applications.
The production module covers human-in-the-loop approvals, retries, idempotency and rate limits, structured outputs, and how each platform bills (tasks, operations or credits, executions, flow actions) on top of model tokens. The governance module covers prompt injection through emails, forms and documents that automations read, secrets and connector permissions, data loss prevention policies, and an inventory and change-control process for citizen-built automations. Platform capabilities and billing units are as of October 2026. Completion and a passing exam score earn a Wirewalk Institute of AI certificate of completion and assessment; it is not a vendor certification or endorsement.