Protect your data
Which tools are approved, what can and cannot go into them, why personal accounts are a risk, and how company accounts keep data out of model training.
Workshop · For everyone who uses AI at work
Using AI Safely, Well and Accurately
A practical session that turns occasional users into confident ones: what you can and cannot put into AI, how to get answers you can rely on, and the techniques power users use every day.
- Length
- 2 hours, or a series of lunchtimes
- Group
- Up to 30, any role
- Format
- On site or remote, laptops open
- Tools
- Claude, ChatGPT, Copilot
Three things every user needs to get right.
Most AI risk and most wasted effort come from the same few habits. Fix those and adoption takes care of itself.
Get better results
A six-step way to brief AI, reusable instructions, projects and skills, and how to turn a one-off success into something the whole team can use.
Trust, but verify
Why AI gets things wrong, how to ask for sources and check them, when to keep a human in the loop, and how to spot a confident wrong answer.
The agenda
Two hours, mostly hands-on.
- 0:00 15 min
Where AI helps, and where it goes wrong
Real examples of AI saving days of work, and real examples of it inventing facts, leaking data and producing off-brand content.
- 0:15 20 min
Safely: the data rules
Approved tools and accounts; customer, employee and financial data; confidential documents; and a simple traffic-light test for what goes in.
- 0:35 35 min
Well: briefing AI like a power user
The six-step prompt (role, task, context, examples, format, checks), then hands-on practice on each person's own work. Saving what works as reusable instructions, projects and skills.
- 1:10 25 min
Accurately: checking the output
Asking for sources and following them, cross-checking numbers, working from your own documents rather than the model's memory, and when a human must sign off.
- 1:35 25 min
Sharing and next steps
How to share good tools with colleagues, champions and lunchtime sessions, the company AI policy in six points, and questions.
The six-step prompt
A better brief gets a better answer.
The single habit that most improves results. Everyone leaves with a one-page guide and their own examples.
- 1RoleWho should the AI be?
- 2TaskWhat exactly do you need?
- 3ContextBackground, audience, documents.
- 4ExamplesShow what good looks like.
- 5FormatLength, structure, tone.
- 6ChecksAsk it to cite sources and flag doubts.
What everyone leaves with
A one-page AI policySix points everyone can remember, adapted to your company.
The prompting guideThe six-step brief, with worked examples.
A checking routineHow to verify sources, numbers and claims before you rely on them.
Their own toolA saved, reusable assistant for one real task in their job.