AI Guides › Playbooks
By Nigel Guy · 5 min read
Most people collect free AI certificates the way they collect browser tabs: enrol in everything, finish a few, and end up with a folder of PDFs that nobody asked to see. It feels productive because every course has a progress bar. It fails because a completion certificate from a free course is a record that you watched and passed a quiz, not proof that you can do anything at work.
The rule: pick one credential for one job you will do in the next 30 days, finish it, then use it on real work before you enrol in anything else.
A ledger is a short table you fill in before you enrol. One row per candidate, four columns. If you cannot fill a column, you have not read the course page properly.
| Column | What you write | Why it matters |
|---|---|---|
| Job | The specific task you will do differently afterwards | Stops you collecting for its own sake |
| Real cost | Money, plus hours, plus any trial that converts to a subscription | "Free" often means free to start |
| What the paper says | Completion certificate, or a proctored credential | These are not the same thing |
| Proof | The piece of work you will have made by the end | This is what you can actually show someone |
This is what we could verify on official pages on 2026-10-04. Course lists change often, so check each page before you commit.
| Option | Cost at time of writing | What you get | Catch |
|---|---|---|---|
| Anthropic Academy (Claude Academy / Skilljar) | Free; sign-up only | Self-paced courses from beginner (Claude 101, AI Capabilities and Limitations) to developer (Claude API, MCP, Claude Code). The AI Fluency course lists 14 lessons and about 4 hours. | Mostly Claude-specific. Certificates are described as completion certificates, not exams you sit under supervision. |
| OpenAI Academy | Free to anyone with a ChatGPT account, per OpenAI's help centre | Courses such as AI Foundations (60 to 75 minutes) and Applied AI Foundations, with course-completion certificates | OpenAI states these are not OpenAI Certifications and do not guarantee eligibility for one later. |
| Google AI Essentials | Google's page says free; the Coursera listing says $49 a month after a 7-day trial (about £37 at time of writing, check the £ price at checkout) | About 5 hours of content on Google's page; completion certificate | The two pages describe cost differently. Treat it as free only if you finish inside the trial or qualify for financial aid. |
| Google AI Professional Certificate | Same $49 a month after a 7-day trial (check the £ price at checkout) | Longer programme, suggested 3 to 6 months, with 3 months of Google AI Pro included | Not free in any sustained sense. A fast learner may finish inside the trial; do not plan on it. |
| Microsoft Azure AI Fundamentals | Learning material is free; the proctored exam AI-901 is paid, priced by country (check the exam page) | A real, proctored Microsoft credential | Aimed at aspiring AI solution builders; the exam page expects some Python and Azure familiarity. |
Two things in that table trip people up. First, the Microsoft route is the only one here that ends in a supervised exam, and that exam is the part that costs money. Second, the exam is now AI-901; older articles still talk about AI-900.
Priya runs a three-person bookkeeping practice and wants to cut the time she spends on client chasers and monthly summaries. Her ledger:
She rejects the Google professional programme (months, subscription) and the Microsoft exam (wrong role). One short course, one artefact. That is a finished ledger.
Go deeper in the tool you actually use, not in more certificates. The Anthropic catalogue goes into API, MCP and agent skills; the Microsoft route goes into Azure services. Depth means a build with your own data, not a longer list of badges.