# Recommended Learning Sequence
Which OpenAI track to take, in what order, for each role – with time budgets, prerequisites and a four-phase plan from zero to the developer paths.
import { Card, CardGrid, Tabs, TabItem } from '@prosefly/astro-components';
There is one sequence that works for almost everybody, and a few role-specific detours. The sequence matters because each track assumes the previous one: agents make no sense before workflows, and workflows make no sense before prompting and verification.
## The default sequence
```text
┌───────────────────┐
│ AI Foundations │ prompting · context · verification · responsible use
└─────────┬─────────┘
▼
┌───────────────────┐
│ Applied AI │ decomposition · contracts · capability choice · review points
└─────────┬─────────┘
▼
┌───────────────────┐
│ Agents & Work- │ objectives · boundaries · oversight · reliability
│ flows │
└─────────┬─────────┘
│
┌───────────┴───────────┬────────────────────┐
▼ ▼ ▼
┌───────────┐ ┌────────────┐ ┌──────────────┐
│ API path │ │ Codex path │ │ AI Leadership│
│ (builders)│ │ (engineers)│ │ (owners) │
└───────────┘ └────────────┘ └──────────────┘
```
AI Leadership is deliberately not gated: an executive can take it first, but it lands far better after AI Foundations because governance conversations depend on knowing what the tools actually do.
## Choose by role
**Sequence:** AI Foundations → Applied AI Foundations → Agents and Workflows.
You are the primary audience for the Foundations pathway. Your work is prompting, context management, verification and turning ad-hoc AI use into workflows your team can repeat. Budget 12–16 hours including our mock exams, and do the hands-on exercises in a real ChatGPT account rather than reading them.
**Skip:** the API and Codex tracks. Read the [ChatGPT feature reference](/appendix/openai/chatgpt-features/) instead when you need product detail.
**Sequence:** AI Foundations → Applied AI Foundations → AI Leadership → Agents and Workflows.
You need enough hands-on fluency to tell a good pilot from a demo, then the governance and adoption vocabulary. AI Leadership is the longest single Academy course (180 min) and the one where our leadership track's scenario work pays off most. Budget 18–22 hours.
**Sequence:** AI Foundations (fast) → Agents and Workflows → API Developer Path → Codex Path.
Do not skip the non-technical tracks entirely: the API assessments test scoping and evaluation judgment, which is exactly what the Applied AI and Agents material teaches in plain language. Then go deep on the API track – it is the largest one here, covering all five Academy API courses. Budget 25–30 hours.
**Sequence:** AI Foundations (fast) → Codex Path → API Developer Path (D4 and D7 at minimum).
The Codex pathway is about scoping tasks, verifying changes, evidencing review, and then scaling that across a governed team. If you own the rollout, read the API track's production-safety domain too. Budget 15–20 hours.
**Sequence:** AI Foundations → Applied AI Foundations, then the standalone Academy courses **AI for Educators** or **AI for College Students** (45 min each).
Those two courses are not mirrored as tracks here because they are short and highly context-specific; the Foundations track covers the underlying skills they assess, and the [credential landscape](/openai/credentials/) page covers ChatGPT Foundations for Teachers.
## Time budget
| Track | Academy courses mirrored | Academy time | Our material (study + 2 mocks) |
| --- | --- | --- | --- |
| AI Foundations | AI Foundations | 60–75 min | 5–7 h |
| Applied AI Foundations | Applied AI Foundations | 75–90 min | 4–6 h |
| Agents and Workflows | Agents and Workflows | 75–90 min | 4–6 h |
| API Developer Path | Scope AI Solutions · Evaluate AI Applications · Design and Build Agentic Systems · Build with RAG · Optimize AI Application Performance | 340 min | 10–14 h |
| Codex Path | Get Started with Codex · Extend Codex Workflows · Scale Codex Across Governed Teams and Systems | 270 min | 8–11 h |
| AI Leadership | AI Leadership | 180 min | 5–7 h |
The ratio is intentional: Academy courses are short because they are guided. Independent preparation costs more time because you are also building recall and judgment that survive a randomised assessment.
## How to work a single track
1. **Read the track overview.** It lists the domains, weights, and what the assessment really tests.
2. **Work the domain pages in order.** Each has a mental model, a decision framework, worked examples, common mistakes, a scenario challenge, and 10–24 practice questions. Answer before expanding.
3. **Do the hands-on checklist** on the track overview in a real ChatGPT, API or Codex environment. Every track has one. Reading about capability selection teaches you nothing about capability selection.
4. **Sit mock exam 1 untimed** as a diagnostic. Note your two weakest domains.
5. **Re-read those two domains**, then take the Academy course itself.
6. **Sit mock exam 2 timed.** Reach 80%+ before you take the Academy assessment; reach 90%+ before you move to the next track.
## Four-phase plan if you are starting from zero
| Phase | Focus | Outcome |
| --- | --- | --- |
| 1 | AI Foundations | You can write a clear prompt, supply the right context, and verify an output before you use it |
| 2 | Applied AI Foundations | You can turn a recurring task into a documented, repeatable workflow with review points |
| 3 | Agents and Workflows | You can delegate a multi-step task to an agent with explicit boundaries and check its work |
| 4 | API and/or Codex | You can build, evaluate and operate an AI application, or ship code with an agent under governance |
AI Foundations · 7 domains
Before your first mock