By invitation Access by access key.
132 topics. 89.6 hours. A shared core for everyone, a track for your job, and a Prompt Lab that scores what you write out of 100 and tells you exactly what is missing.
Access is by invitation. Sign in with the Google or Microsoft account you already have, then enter the access key an administrator issues you.
Same six checks for everyone. Plain code, not a model guessing — the same prompt always gets the same score.
Why this exists
You can finish ten hours of AI content and still freeze in front of an empty prompt box. Passive courses measure attendance. This one measures work.
The curriculum
Five phases over about 17 weeks. Everyone starts in the same place; the track you add depends on whether your output is code, documents or decisions — or which cloud and hardware you build on.
Everyone takes the same core: 40 topics · 24.7 h Foundations, prompting, model architectures, deployment and responsible AI — in plain English, whatever your role. Then you add one of these.
Everyone — the fastest payback for any role
12Topics
5.8Hours
The habits that separate a vague request from a prompt that returns work you can actually use. Additive — open to all, whichever other track you pick.
Example topics
With the core: 52 topics · 30.4 h
Developers, testers, architects, data engineers
20Topics
16.4Hours
From calling an LLM API to shipping something you would put in front of customers: structured outputs, RAG, agents, evaluation, tracing, cost and security.
Example topics
With the core: 60 topics · 41.1 h
PMs, analysts, admins, marketing, ops, leadership
20Topics
11.8Hours
Turn AI into an everyday work habit — drafting, summarizing, analysing research and spreadsheets, role playbooks, no-code automation and the privacy rules that keep you safe.
Example topics
With the core: 60 topics · 36.5 h
Engineers, architects and analysts working on AWS
10Topics
7.8Hours
The AWS generative AI stack end to end: foundation models through Amazon Bedrock, custom models on SageMaker AI, RAG with Knowledge Bases, agents on AgentCore, and the guardrails a security review will ask about.
Example topics
With the core: 50 topics · 32.4 h
Developers, data engineers and IT professionals in Microsoft shops
10Topics
7.6Hours
Microsoft's AI platform as it stands today: Microsoft Foundry and its model catalogue, grounding with Azure AI Search, prompt and hosted agents, Content Safety, and Entra-based governance.
Example topics
With the core: 50 topics · 32.2 h
Engineers, analysts and ML practitioners on Google Cloud
10Topics
7.7Hours
Google Cloud's AI surface: Gemini and Model Garden, Vertex AI Vector Search, grounding and citations, the open-source Agent Development Kit, BigQuery ML and responsible AI controls.
Example topics
With the core: 50 topics · 32.3 h
Platform engineers, ML engineers and anyone self-hosting models
10Topics
7.9Hours
One layer below the managed clouds: how GPUs and VRAM constrain AI, deploying models as NIM microservices, training and fine-tuning with NeMo, self-hosted RAG, and programmable safety with NeMo Guardrails.
Example topics
With the core: 50 topics · 32.6 h
The Prompt Lab
You write a prompt for a real work scenario. Six fixed checks score it out of 100 — published rules, the same for everyone, run in plain code rather than a model's opinion. Then you see the stronger version and where every point went.
Exercise: Write a Leadership Status Update. Your VP reads about 15 of these a month and skims each one in under a minute, so any line that does not change a decision gets cut.
14
Basic
Write a status update for my project.
Six-dimension rubric
Role & context0/20
Task clarity14/20
Specificity & constraints0/20
Format & structure0/15
Audience & tone0/15
Examples & evidence0/10
100
Expert
You are a project lead writing the monthly update for your VP. Context: She reads about 15 of these a month and skims each one in under a minute, so any line that does not change a decision gets cut. My raw notes for the month are pasted at the end. Task: Write the monthly status update for this project from those raw notes. Constraints: - No more than 200 words in total, with 2 to 3 bullets under each heading. - A severity label (High/Medium/Low) on every risk, and a deadline on every ask. - Avoid filler openers such as "as you may know", and do not report progress that is not in the notes. Output format: three bold headings — Progress, Risks, Asks — each with short bullets, then a "Bottom line" of no more than three sentences. Audience and tone: written for one busy VP rather than the wider team; direct, factual and unpolished, in plain English with no jargon. Grounding and example: use only the notes below, and where a number is missing say so instead of estimating. A good Progress bullet reads: "Shipped the redesigned onboarding flow; early data shows 18% faster time-to-first-action for new users."
Six-dimension rubric
Role & context20/20
Task clarity20/20
Specificity & constraints20/20
Format & structure15/15
Audience & tone15/15
Examples & evidence10/10
10 Prompt Lab exercises 6 scored dimensions 4 bands: Basic → Developing → Strong → Expert Every exercise ships with a weak sample and a strong sample, so you can see the gap before you close it.
What's inside
No marketing rounding. These are the actual contents of the platform today.
132Topics across all seven tracks
89.6 hOf content if you did everything
396Quiz questions
57Glossary flashcards
5Workbooks
6Hands-on projects
AI instructorAsk questions on any topic, in context
Progress trackingHours, quiz scores and certificates, recorded
Structured as 5 phases over about 17 weeks — take it at that pace or go faster. Your progress is saved either way.
How it works
There is nothing to install and nothing to configure — you just need an access key.
Use the Google or Microsoft account you already have, then enter your access key. No new password to invent.
Prompt Mastery, Engineering, Business & Builder, or one of the cloud and hardware tracks — AWS, Azure, Google Cloud and NVIDIA. The 40-topic core is yours either way, and you can change track later.
Read the lesson, take the quiz, work the workbook, build the project, then score a prompt in the Lab. Ask the AI instructor when you get stuck.
Hours, quiz scores, prompt scores and certificates are recorded as you go, so improvement is a number rather than a feeling.
Getting in
Access to the courses is by invitation. Sign in with your Google or Microsoft account, then enter the access key an administrator gives you. If you do not have a key yet, contact your administrator to request access.
Sign in with your Google or Microsoft account, then enter the access key your administrator issues you.
Google Microsoft Already have an account? Sign in.