Team enablement taught on your own systems and your own documents.
Hands-on sessions inside your own systems, with your team's real documents. The last session covers when to leave the model out, and its rules go into a playbook the team keeps.
AI training for employees sticks when it is taught on the team's own work, inside the tools they already use, with a clinic two weeks later when the real questions surface. Each person leaves with a few tasks they now do differently, and the team keeps a written playbook of prompts, checks and the lines a model must not cross.
Training sticks when the examples are your own.
A marketing prompt demo does nothing for an operations manager who spends Tuesdays reconciling delivery notes, and an hour on model architecture helps almost nobody do their job on Monday.
Your team is probably using these tools already. In Microsoft and LinkedIn's 2024 Work Trend Index, a survey of 31,000 people in 31 countries, 75% of knowledge workers said they use AI at work, 78% of those users brought their own tools, and only 39% had received AI training from their company. The gap is in using them well, on the right data.
So the curriculum is built after we have seen the work, from your documents, systems, tone and constraints, and by the end each person has three or four tasks they now do differently, written down where they do them.
A six-part spine, adjusted to the team.
- Foundations. What these systems are and are not, what a hallucination is, and why the same question asked twice gets two answers.
- Prompting that survives contact. Structure, context, examples and constraints, practised on your documents until it is habit.
- Your own tools. Whatever we have built for you, plus the ones you already pay for, used hands-on.
- Checking an answer fast. Citations, spot checks and the two-minute rule before anything reaches a client.
- Where the line is. Money, legal text, first contact with a customer and anything irreversible, written down against your organisation's own rules.
- A playbook that stays. One short internal document with the prompts, the rules and the escalation path, so the next hire gets it on day one.
The audience decides the format.
Programmes run from one half-day to six weeks, delivered remotely.
| Who | What they practise | Format |
|---|---|---|
| Operations, finance and support teams | Prompting on their own documents, quick answer checks, and the lines a model must not cross | Half-day blocks, then a clinic two weeks later |
| Managers and owners | Where a model pays in their area and where it does not, and what to demand of any such system before trusting it: an evaluation set, an audit trail and a person on the irreversible steps | A half-day session |
| Technical teams | Retrieval design, evaluation, agent tooling, cost control and failure modes | A separate course over several sessions |
An accounts payable session, step by step.
Worked example. Not client data. An accounts payable team of six spends much of its week matching supplier invoices against delivery notes and chasing the differences.
- The team pulls the fields out of a badly scanned invoice. The three that matter most are checked by hand before anyone trusts the rest.
- Each person asks the model to list every mismatch between an invoice and its delivery note, then compares that list with their own.
- The supplier query is drafted with the model, in the team's own tone, and a person sends it.
The rule for their playbook is short: the model drafts, a person checks and sends, and no payment is released on the model's word. At the clinic two weeks later we check which habits held, and why.
Five steps, from scoping call to playbook.
Nothing is locked inside a course platform.
- Scoping call. The people in the room, their work, what they already try with these tools, and anything that has gone wrong.
- A curriculum built on your work. We ask for a sample of a normal week's work, redacted where needed, and turn it into exercises.
- Sessions. Half-day blocks, held remotely in small groups, with everyone on their own machine.
- A clinic two weeks later, for the questions that only appear once people have tried, which is where habits stick or fail.
- Playbook and handover. The written rules and prompts, in your hands and editable.
The last session is about leaving the model out.
It covers the confident wrong answer, the decision a regulation reserves for a person, and the formula that is cheaper and exact, so the team also learns when the rest can be trusted.
Our systems follow the same rule. Spending money, deleting records and first contact with a customer are blocked for the model in code, and the training teaches that line to people, so the tools and the team agree.
Human oversight is one of the twelve principles in the UAE Charter for the Development and Use of Artificial Intelligence, alongside promoting AI awareness. Article 4 of the EU AI Act asks providers and deployers of AI systems to take measures to support the development of AI literacy of their staff, in the wording set by Regulation (EU) 2026/1744, in force since 27 July 2026. We give no legal advice on either.
The team keeps a playbook and changed habits.
SolaryxDev holds no training accreditation and issues no certificates or CPD points. The team keeps the tasks each person now does differently and the prompts that worked, in a document it owns and edits.
The playbook is short on purpose. It names the tools the team may use, which kinds of data may go into each, the checks before anything leaves the building, and who to ask when an answer looks wrong. When your systems or rules change, it is the first document the team updates.
Frequently asked questions
What does team training involve, and what makes it stick?
AI training for employees sticks when it is taught on the team's own work, inside the tools they already use, with a clinic two weeks later when the real questions surface. Each person leaves with a few tasks they now do differently, and the team keeps a written playbook of prompts, checks and the lines a model must not cross.
Who should be in the room?
The people who do the work, and their manager. Mixed groups work best, because the person who knows the exception is the one the exercises need, whatever their seniority.
What does team training cost?
One fixed price, set in the proposal after the scoping call. It depends on the number of sessions and people: a single half-day block for one team is the smallest programme, and six weeks the longest.
Which language are the sessions taught in?
English. The exercises can still use your Arabic or mixed-language documents, which the systems we build read.
What if we have not built anything yet?
Then training is premature for most of the team. Start with a short session for two or three people alongside a strategy engagement, and train the wider team once there is something to use.
Do participants get a certificate?
No. We are not an accredited training provider. People leave with work they now do differently, and the team keeps the playbook.
Can the exercises use our confidential documents?
Yes. We work from a redacted sample of real work, under your NDA or ours, and nothing you give us trains a model outside your company. Part of the session is deciding which tools your material may enter, and that decision goes into the playbook.