✦ Velza Academy · AI Workflow Engineering
Three weeks of live, hands-on work on the full AI engineering stack. Prompt and context engineering, memory and state, sub-agents and agent teams, loop engineering, and the harness work that makes AI reliable. Built for non-coders.
✦ Your instructor
Mike Wheeler is an AI and Salesforce trainer and O’Reilly author. He has taught well over 500,000 global learners on major platforms such as edX, LinkedIn Learning, Pearson, and Udemy, and he built the first generative AI course on edX. He teaches for non-coders, working backward from your career goals.
Professionals who want to genuinely operate AI systems, not just use them. If you can use a browser and a spreadsheet, you can do this.
Most people can prompt. Far fewer can build an AI workflow that holds up when it matters. This cohort is about the difference: how context, memory, loops, and the harness around a model decide whether it works once or works every time. No coding required.
✦ The curriculum & schedule
Six live sessions over three weeks: three teaching sessions plus three optional office hours. All times Central. Can’t make one live? Every session is recorded.
How to structure prompts that actually work, and the system underneath them: models, context windows, and what belongs in each.
Job-market intel and open Q&A, shared across every Velza Academy cohort.
Where information lives across steps (prompt, file, database, summary), sub-agents and agent teams, and how controlled loops observe, decide, act, verify, and stop.
Job-market intel and open Q&A, shared across every Velza Academy cohort.
Agent equals model plus harness. Guardrails, verification, blueprints, and how to attribute a failure and fix the harness rather than just the prompt.
Job-market intel and open Q&A, shared across every Velza Academy cohort.
How to structure prompts that actually work, and the system underneath them: models, context windows, and what belongs in each.
Job-market intel and open Q&A, shared across every Velza Academy cohort.
Where information lives across steps (prompt, file, database, summary), sub-agents and agent teams, and how controlled loops observe, decide, act, verify, and stop.
Job-market intel and open Q&A, shared across every Velza Academy cohort.
Agent equals model plus harness. Guardrails, verification, blueprints, and how to attribute a failure and fix the harness rather than just the prompt.
Job-market intel and open Q&A, shared across every Velza Academy cohort.
How to structure prompts that actually work, and the system underneath them: models, context windows, and what belongs in each.
Job-market intel and open Q&A, shared across every Velza Academy cohort.
Where information lives across steps (prompt, file, database, summary), sub-agents and agent teams, and how controlled loops observe, decide, act, verify, and stop.
Job-market intel and open Q&A, shared across every Velza Academy cohort.
Agent equals model plus harness. Guardrails, verification, blueprints, and how to attribute a failure and fix the harness rather than just the prompt.
Job-market intel and open Q&A, shared across every Velza Academy cohort.
✦ What you’ll walk away with
Companies are adopting AI faster than they can find people who know how to operate it.
Study after study points to the same gap. The professionals who can prompt, shape context, and build reliable AI workflows are the ones getting hired and promoted. This is where you start.
✦ What you’ll learn
This is a skill track, not a certification. You will build fluency across the full engineering stack:
| Prompt Engineering |
| Context Engineering |
| Memory & State |
| Sub-agents & Agent Teams |
| Loop Engineering |
| Agent Harness |
| Blueprints |
✦ What’s included
Three teaching sessions plus three optional office hours, live over three weeks. Sundays at 7:30-8:30 PM Central.
Every session is recorded, and you keep access to the replays for a full year, so you can rewatch any time you need them.
A real workflow you build and break each week, so you leave with something that runs rather than notes.
✦ Questions, answered
No. Everything in this cohort is done without programming. We work with the tools and interfaces you already have, and the engineering is in how you structure the work, not in syntax.
No, and we will not pretend otherwise. AI workflow engineering has no vendor exam behind it. This is a skill track: you leave able to do the work, with things you have built to show for it.
Prompting is one of five layers we cover. The other four (context, memory, loops, and the harness) are what separate an AI workflow that works once from one that works every time.
We stay tool-agnostic on purpose. The concepts transfer across whatever you or your employer already run. Bring the AI tool you actually use.
Every session is recorded and shared, and you keep access to the replays for a full year, so you can watch or rewatch any of them on your own schedule. The optional Friday office hours are recorded too.
Self-study stalls. A live cohort gives you a fixed three-week pace, the chance to ask questions in real time, weekly hands-on work, and optional office hours. You actually finish.
You can request a full refund within 24 hours of the start of the first live session.