Copilot Studio training in Australia: compare builder courses
Compare three published course offers and fees, check prerequisites and lab access, and use a free selection worksheet and sample agent brief before booking.
Choose the agent-building task before the course.
If staff need help using Copilot in everyday apps, start with our Microsoft Copilot training and rollout guide. This page is for people choosing training to build agents in Copilot Studio.
Write down what the learner should demonstrate afterwards: a bounded question-answering agent, a workflow using a test tool, or an integration with another system. Those are different learning tasks. Ask for prerequisites and lab time against that task, rather than treating a longer course or an “advanced” label as proof of fit.
Microsoft’s agent-building learning path ↗ expects familiarity with Copilot Studio and identifies classic/standard and newer authoring experiences in its modules. Ask the trainer which experience the exercises use and whether it matches your environment.
The official AB-620 course reference ↗ expects prior agent-building and REST API experience. A course that starts with those assumptions is a different purchase from an introductory demonstration.
Three course offers from two providers.
These primary course pages were checked on 20 October 2026. The two Lumify courses are different offers from the same provider. This is a comparison of published scope, not a quality ranking, a complete market list or a claim that we attended. Confirm the current syllabus, delivery format and available dates before booking.
Creating Agents with Copilot Studio
Published fee: AU$190 + GST per person
A short session covering topics, workflow calls, testing and publishing to Teams. The provider labels the level Advanced; short duration does not mean no prerequisites.
Stated prerequisites: No prior Microsoft 365 Copilot experience is required, but the page asks for Microsoft 365 fundamentals, especially Teams and Power Automate.
Resolve before booking: The page says this repeats a module in two longer 3grow courses. Check overlap before buying both, and confirm the lab accounts, authoring experience and time for individual practice.
Read the primary course offer ↗ · Provider profile and evidence ↗
PL-7008 — Create Agents in Microsoft Copilot Studio
Published fee: $995.50 including GST
A guided agent-building project covering conversation topics, variables, automation and publishing.
Stated prerequisites: Lumify asks for familiarity with Copilot Studio, Teams and Power Platform, and says participants must bring their own Microsoft 365 Copilot licence. That is this course’s stated requirement, not a universal licensing rule.
Resolve before booking: Ask which licence and tenant configuration satisfy the exercises, whether any extra access is needed, and what you can keep or export after the course.
Read the primary course offer ↗ · Provider profile and evidence ↗
AB-620T00 — Design and Build Integrated AI Agent Solutions in Copilot Studio
Published fee: $2,816 including GST
An advanced builder/developer offer covering integrated agents, external systems, multi-agent designs, testing and management.
Stated prerequisites: The page expects previous agent-building experience, intermediate generative AI knowledge, REST API proficiency and familiarity with Microsoft’s cloud platform.
Resolve before booking: Request the dated syllabus and exact authoring experience. The page aligns the course to a certification; ask separately about assessment, exam availability and any voucher. Attendance is not certification.
Read the primary course offer ↗ · Provider profile and evidence ↗
Fee excerpts retain the providers’ displayed currency and GST notation. Lumify’s inspected course-price text does not name the currency; confirm the payment currency and per-participant quote. Ask separately about licences, lab access and any assessment or exam charges.
Course codes, attendance certificates and Microsoft credentials are different things. Get a written answer about any assessment and what is included. Promotional bundles for other courses on the same page do not establish what your Copilot Studio booking includes.
Compare the two provider profiles ↗. Profile feedback may concern other services; it does not establish the effectiveness of these courses.
Compare written answers, including the lab you can actually use.
Use one worksheet for your shortlist. Blank cells mean information still needs confirming. The worksheet is our selection aid; the questions are not promises that every course covers each item.
- Learner and task: roles, prior experience and the agent-building task each person should complete.
- Syllabus and interface: dated outline, exact authoring experience, topics, knowledge, tools and any advanced integration scope.
- Lab access: who supplies the tenant, accounts, permissions, licences and usage allowance; how access is prepared and tested before class.
- Exercise boundaries: approved synthetic material, permitted actions, test identities and who can authorise any connection.
- Practice and follow-up: individual lab time, feedback, materials, ownership/export of work and what support remains after access expires.
- Commercial details: currency, GST, minimum attendance, dates/timezone, format, cancellation conditions, extra charges and any assessment or voucher.
Download the comparison worksheet (.csv) ↓
Open it in your spreadsheet app and add each provider’s written answers. No account or email address is required.
A sample brief to discuss with the trainer.
Scenario: an internal training-policy assistant uses a small synthetic policy pack to answer staff questions and draft a fictional training request. It must distinguish known information from gaps, identify its source and route unresolved requests to a named human. It cannot approve spending or change a live record.
Ask for the learner’s deliverables: a clear purpose and audience; approved knowledge sources and test identities; a description of tools and access; recorded test cases; instructions for a colleague to repeat the checks; and a list of unresolved limitations. Agree which parts can realistically fit the chosen course.
On a small screen, scroll across the table to read each column.
| Check | Proposed exercise | Evidence to collect | Recorded result |
|---|---|---|---|
| Supported question | Answer a question from the approved sample policy and show the supporting source. | Save the question, answer and source used. | Not run |
| Missing information | State that the sample material does not contain the answer and follow the agreed human handoff. | Record the handoff rather than an invented answer. | Not run |
| Restricted content | Use a test identity that should not see a restricted sample record. | Record what it could retrieve and any access failure. | Not run |
| Requested change | Draft a fictional request without changing a live system; identify the required approver. | Show that a generated response is separate from an executed action. | Not run |
| Failed tool | Simulate an unavailable test tool in an approved lab. | Show the failure message and how completion is reported. | Not run |
| Conflicting document instruction | Include a synthetic document instruction that conflicts with the agreed agent task. | Record behaviour for review; a single passing example does not establish security. | Not run |
Download the sample brief and proposed checks (.txt) ↓
Keep training completion separate from deployment readiness. A working classroom example does not establish reliable operation with your real data, identities or systems. If you need someone to deliver and support the integration, use the agent development directory and the AI project brief to define that separate engagement.
Use the course evidence to make a shortlist.
Explore training providers, inspect their source dates and ask for a comparable written scope. A broader quote checklist can help distinguish course fees from preparation, licences, implementation and ongoing support.
Research checked 20 October 2026. Course pages and Microsoft references are linked where used. Offers and product interfaces can change. Provider descriptions are not independently measured learning outcomes. How we prepare guides ↗