Using AI Safely at Work
Your staff are already using AI tools. This short course is built around your business: the tools you allow, the information you hold, and the person they should tell when something goes wrong. You get a record you can show.
It's the cover-up that costs, not the cock‑up.
Mistakes with new tools are normal. The expensive one is the mistake nobody mentions until a customer or an insurer finds it. So every scenario in the course comes back to one habit: say so straight away, to a named person, in a way they can act on.
Four steps, most of them ours
Tell us about your business
A 15-minute online form: your sector, the AI tools in use, the information your staff handle, and who they should tell when something goes wrong.
We build your course
Your rules in plain English, scenarios from your sector, and questions that use your tools and your named person. You check it before anyone sees it.
Staff take it
Each person gets a personal link. A one-time code sent to their work email confirms it's them, so the record shows the right person took it.
You get the record
A live page showing who has finished, their score and the date, plus a spreadsheet download and a summary of what your team found hard. Completions can also go straight to your HR system or CRM.
Two samples from the course
Both come from a course built for a fictional haulage firm, Hartley Freight. In a real course, the names, tools and rules come from your answers.
The rate card is one of the three scenarios: pick an option and see where it leads. Spot the mistakes asks staff to check an AI-written email against the booking it came from. Nothing you choose here is recorded.
What your business gets
- Your rules, in plain EnglishFive rules built from your answers. These double as a one-page AI use guide for staff.
- Scenarios from your sectorRealistic situations, written as message threads, with a wrong-turn branch for each choice.
- A voicemail that isn't what it seemsAn urgent message that sounds like a senior colleague, asking for something they shouldn't. Staff decide what to do before they find out.
- Spot the mistakesAn AI-written email or letter from your sector with three errors in it. Staff check it against the source, the way they should before anything goes out.
- Answers in their own wordsThree "what would you do?" questions, answered in writing and marked against a checklist we set, with feedback straight away.
- Practice at owning upStaff write the message they'd send to your named person after a mistake, and get coaching until it's clear, quick and complete.
- Which AI tools are really in useAn optional anonymous question. You see totals only, once at least five people have answered, so staff can be honest.
- Questions with a pass markTen questions marked on our server. Staff can retake them; the record shows the final result.
- A signed acknowledgementEach person confirms the rules by typing their name at the end.
- Records that can't be quietly editedEvery step is logged in order, and the report checks the log is complete each time it opens.
- What your team found hardA summary for managers: which scenarios caught people out, which written answers fell short, and the tools in use. Anonymous, and only shown for groups of five or more.
- A yearly refresherEach record shows when the refresher is due, and small updates are included when your tools change.
A course about AI should handle AI carefully
The written answers and the owning-up practice are marked by AI, using Anthropic's Claude, against a checklist written by us. Staff are told this before they start.
- Only the answer is sentThe question, our checklist and what the person wrote. Not their name, email or your organisation.
- Not kept, not used for trainingAnthropic doesn't keep this by default and doesn't use it to train its AI.
- AI doesn't decide who passesThe pass mark comes from the ten questions. Every written answer and its feedback is in the record for you to read.
- Staff are told not to use real namesThe course reminds them before every written answer.
Organisations of 10 to 249 people
Big enough that staff are using AI in ways nobody has written down, too small to have a training department. There are scenarios ready for:
One price for the year, by number of staff
The price depends on how many people will take the course. Book a short call and we'll give you a fixed price for the year.
The first year includes building your course. Every year includes hosting, records, the refresher, and small updates when your tools or contacts change. Larger changes during the year, such as new scenarios or a change in regulation, are quoted at a fixed price before any work starts. There's no VAT to add: DTTtech isn't VAT registered.
What this is, and what it isn't
This is practical training that helps your staff use AI tools sensibly and speak up quickly when something goes wrong, with a record of who has taken it. It isn't legal advice, and it doesn't on its own make a business compliant with any law or meet any particular insurer's requirements. If you need that, we're happy to work alongside your adviser.
Book a short phone call
Fifteen minutes to hear how your team uses AI and whether this would help. No obligation, and if it isn't a fit I'll say so.
Michael O'Sullivan, DTTtech · michael@dtttech.com
IT support and managed service providers: we also build the course for your clients, under your name or theirs. Mention it when you get in touch.
DTTtech is a small, independent eLearning company in Fleet, Hampshire. Michael O'Sullivan has spent ten years building training for clients in financial services, logistics, healthcare and other regulated sectors.
If you've been asked to take the course by your employer, use the personal link in your invitation email.