What does AI training for a business cover?
Good AI training teaches your team to use AI on the work they already do, safely, and to check what it gives back. I run AI training Australia wide as in-house workshops, in your office, built around your team's real tasks rather than generic demos.
A session can cover:
- Prompting for real tasks: quotes, emails, reports, meeting notes, tenders and procedures, using your own examples.
- Simple assistants and projects: setting up a ChatGPT or Claude project with your documents and instructions, so the team stops pasting the same background in every day.
- Safe use with client data: what can go in, what never goes in, and which plans keep your data out of model training.
- A simple AI use policy: one page your team can actually follow.
- Where AI goes wrong: made up facts, wrong numbers and confident mistakes, and how to check for them.
The aim is that people walk out using AI on Monday morning, not just knowing it exists.
What happens in a half day or full day workshop?
A half day gets a team confident with the basics on their own tasks. A full day adds hands-on building, where each person leaves with a working assistant or project for a job they do every week.
| Format | Who it suits | What you leave with |
|---|---|---|
| Half day team workshop | Teams new to AI, or using it unevenly | Prompting on real tasks, safe use rules and a draft AI use policy |
| Full day team workshop | Teams ready to make AI part of daily work | Everything in the half day, plus assistants and projects built for your own jobs |
| Owners and sole traders | One person running the business, or a small leadership group | The jobs worth handing to AI first, set up on your own work |
Before the day, I ask for a handful of real tasks from the team: an email they write every week, a report, a quote, a procedure. The session is built around those, so nobody spends the day on examples that don't apply to them.
Should your team use Microsoft Copilot, ChatGPT or Claude?
Use the tool that fits where your work already lives. If your team works all day in Outlook, Word, Excel and Teams, start with Copilot. If they need a strong general assistant for writing, research and analysis, ChatGPT or Claude usually does more.
| Tool | Where it works | Good at | Watch for |
|---|---|---|---|
| Microsoft Copilot | Inside Microsoft 365: Outlook, Word, Excel, PowerPoint and Teams | Working with your own emails, files and meetings | Copilot inside the Office apps needs a paid licence, and it can find any file a person already has access to |
| ChatGPT | A separate app on web, desktop and phone | Drafting, research, analysis and custom GPTs for repeat tasks | Free and personal plans handle data differently from business plans |
| Claude | A separate app on web, desktop and phone, plus an API for custom builds | Long documents, careful writing and projects that hold your files and instructions | Same as ChatGPT: check which plan your team is on |
Many teams end up with two: Copilot for the Microsoft 365 work and ChatGPT or Claude for everything else. Whether you book it as Microsoft Copilot training or ChatGPT training for business, the session covers whichever tools you already pay for, and I'll tell you honestly if a tool isn't worth it for your team.
Do you run Microsoft Copilot training?
Yes. Microsoft Copilot training is part of our AI training Australia wide, covered as a tool your team already has in Microsoft 365, focused on using it well on real work in Outlook, Word, Excel and Teams.
As of November 2026, Microsoft includes Copilot Chat at no extra cost with eligible Microsoft 365 business plans, and Copilot inside the Office apps is a paid per-user add-on. Plenty of teams already have the first and don't use it, so that's often where a session starts.
To be clear about what this isn't: Source Digital is not a Microsoft partner and holds no Microsoft certification. Licensing, tenant settings and file permissions are for your IT provider. The training is about getting your people productive with the tool once it's switched on.
Is it safe to put client data into ChatGPT or Claude?
It depends on the plan, and that's the first thing to sort out. As of November 2026, ChatGPT Business and Enterprise, and Claude Team and Enterprise, do not use your inputs to train their models by default. Free and personal plans work differently, and the settings change, so check your own.
Plan choice is only half of it. The other half is habits:
- Put staff on a business plan the business controls, not personal accounts.
- Agree what never goes in, like client identity documents, health details or anything under a confidentiality clause.
- Strip names and numbers out when they aren't needed for the task.
- Keep a person responsible for anything that goes to a client.
This isn't legal advice. If your business handles sensitive personal information, check your obligations under privacy law and any client contracts as well.
Where does AI go wrong, and how do you check its work?
AI gets things wrong confidently. It can invent a fact, a quote or a reference, get a sum wrong, or miss context it was never given. The fix is a checking habit, not avoiding the tool.
- Ask where it came from. Have it quote the source document and page, then look.
- Check every number. Totals, dates, prices and quantities get checked against the source, every time.
- Give it the context. Most bad answers come from a thin prompt. Your documents and examples fix more than clever wording.
- Keep a person signing off. AI drafts. A person approves anything that leaves the business.
It's the same rule I build by. In the tender scorer built for an international tier 1 mining operator, pricing is calculated by deterministic logic and the AI layer has to quote page-referenced evidence for every score.
Who runs the training, and how a workshop is put together
I'm Tristan Sunde, founder and lead developer at Source Digital, and I run every session of our AI training in Australia myself. I build with the Claude API and run Claude and ChatGPT rollouts for teams, and I built the AI systems in our case studies, including a commercial agent in Microsoft Teams that answers staff questions from millions of rows of purchasing and stock data. More about me on my page.
- A short call. Your team size, the tools you use and the jobs that eat the most time.
- A plan and a price. Pricing is quoted per team, in writing, once I know the format and numbers.
- The workshop. In your office, on your team's real tasks, with the tools you already pay for.
- What comes next. If the day turns up a job worth automating properly, that can become a separate build through our AI consulting work. No obligation.
