AI curriculum

Course outline for Workflow Foundations for AI Automation

AI can draft a proposal quickly while leaving a larger queue for human review. Model a request-to-proposal process before deciding which steps belong to rules, models, or people. Make approval and escalation explicit, then compare total handling and review effort rather than model speed alone.

Technology: Workflow Engineering
Duration: 6 hours
Modules: 3 · 2 hours each
Level: Level 1

About Workflow Foundations for AI Automation

Workflow Foundations for AI Automation Course Objectives

Define a case, its inputs, outputs, and completion condition.
Allocate activities to deterministic rules, model-assisted judgment, or human decisions.
Map approval, escalation, waiting, and exception ownership.
Draw a BPMN model that another learner can trace.
Test whether an AI drafting proposal reduces total work or moves it into review queues.

Pre-requisites

Familiarity with a workplace process.
Basic diagramming skills and ability to describe who owns a decision; coding is not required.

Before this course

Complete the required courses or bring equivalent experience. Recommended courses add useful depth; they are not enrolment gates.

No prerequisite course is required. The entry skills above still apply.

Can describe a workplace process and draw a simple diagram; coding is not required.

Find your place in the learning map

Lab Setup

Computer with a text editor, diagramming tool, and spreadsheet.
Synthetic requests, model-draft fixtures, manual/AI-assisted timing records, review-capacity assumptions, and a BPMN reference.
Model capabilities are explained through supplied examples; no AI prerequisite, coding, live model, or production process access is needed.

Detailed Course Outline

Module 1: Allocate work before choosing AI

Understanding work as a system: Connect request inputs, business objects, decision owners, and completion conditions before selecting a bounded automation opportunity.
Activities and tasks: Assign validation to rules, draft generation to model assistance, and approval to a person; justify each allocation from supplied capability assumptions.
Cases: Define the business object that moves through a workflow, including required inputs, outputs, and completion criteria.
Events: Identify occurrences that start or advance a request, distinguishing recorded business facts from planned activities.

Practical outcome: Define a request-to-proposal case and allocate validation, drafting, approval, and sending to rules, model assistance, or people using supplied capability assumptions.

Module 2: Make approval and waiting visible

State: Distinguish generated drafts from approved proposals; specify transitions that prevent incomplete inputs or unreviewed model outputs from advancing to release.
Workflow boundaries: Locate process entry, completion, and team interfaces so exceptions retain a named owner across organizational boundaries.
Handoffs: Name reviewers and escalation owners for model-generated drafts; specify evidence and state needed to continue without assuming the model owns business approval.
Queues: Track requests awaiting model drafting and human review; distinguish faster generation from shorter total waiting time and sustainable reviewer workload.

Practical outcome: Map draft and approved states, reviewer ownership, incomplete-input escalation, and review queues; do not let generated drafts authorize their own release.

Module 3: Test an AI automation hypothesis

Bottlenecks: Compare drafting, review, and correction effort in supplied records; identify whether AI assistance moves the constraint instead of removing it.
Feedback loops: Define total-work measures, review-capacity checks, and scale-or-stop triggers for an AI-assisted process rather than measuring generation speed alone.
Workflow patterns: Recognize reusable sequences, decisions, and exception paths while preserving the responsibilities and completion conditions of the case.
BPMN: Represent activities, events, decisions, and handoffs in a traceable diagram that exposes normal and incomplete request paths.

Practical outcome: Draw a traceable BPMN model and compare supplied manual and AI-assisted timing records, including review, corrections, and queue growth.

Practical exercise

Produce a request-to-proposal model, glossary, and rules/model/human allocation table from a supplied scenario.
Trace normal, incomplete, and uncertain-draft cases; reject a transition that bypasses required human approval and name the escalation owner.
Have another learner identify the next owner and state for all four supplied handoff cases without verbal help.
Compare total handling and waiting time in manual and AI-assisted fixtures; define a review-capacity measure and a scale-or-stop trigger without claiming demonstrated production benefit.

What you will produce

Model an AI-assisted workflow with explicit rule, model, and human responsibilities and a measurable total-work hypothesis.

How the work is checked

Normal and incomplete cases reach documented outcomes through permitted states.
Every handoff names an owner and waiting condition; approval and escalation cannot be bypassed by generated drafts.
A peer correctly traces all four supplied handoff cases.
The proposed improvement includes total handling time, review capacity, and a scale-or-stop trigger rather than model speed alone.

Six-hour noncoding workflow-modeling lab using supplied AI drafts and timing assumptions. Teach BPMN, cases, state, and ownership through one process. Capability assumptions and timing fixtures are not live-model evaluation or proof of business benefit; detailed human-AI allocation follows in later courses.

When another approach is better

A short checklist can be enough for a single-owner process with no meaningful branching.
Do not automate a bottleneck before clarifying its policy and ownership.

How we train

Ask about exercises, instructor feedback and the prior knowledge you need. Tool-specific courses marked provisional may become modules in a broader course.

Discuss your learning goals