Employee AI Training
Rolling out to everybody, without pretending everybody starts in the same place.
What is employee training?
Employee AI training is organisation wide rollout to general staff, where the defining challenge is the range of starting points rather than the content, since one session pitched for everybody is too slow for some and too fast for others.
The difficulty with training everybody is not what to teach. It is that in any group of thirty, several people have been using these tools daily for a year and several have never opened one.
A single session pitched at the middle bores the first group and loses the second, and both leave having gained little. That is why so much organisation wide training produces good feedback and no change.
Streaming by actual starting point rather than by department or seniority fixes most of it, and it costs nothing except asking beforehand.
How we decide who sits where
Duara delivers organisation wide AI training from Nairobi, usually where a tool has been licensed across a business and everybody needs a working level of use.
We stream by starting point rather than by job title, and we ask people beforehand rather than assuming. Seniority is a poor predictor of AI familiarity, and grouping by it puts a director who has never opened a tool next to one who uses it hourly.
Who this is for
- A tool has been licensed across the organisation
- Staff ability varies enormously
- Previous sessions worked for some and not others
- You need a consistent baseline across the business
Industries we serve
- Financial services
- Manufacturing groups
- Retail chains
- Healthcare organisations
- NGOs and development
- Public sector bodies
Problems this solves
These describe rollout at scale, where the range of ability is the real obstacle.
- One session for everybody. Content pitched at the middle bores the experienced and loses the beginners, and both gain little.
- Grouping by department or grade. Seniority predicts AI familiarity poorly, so groups end up as mixed as the whole organisation.
- No baseline afterwards. Everybody attended and nobody can say what the organisation can now reliably do.
- The confident and the silent. A few people dominate the session and the ones who most needed it ask nothing.
What we deliver
Starting points established beforehand, groups streamed by ability rather than grade, content pitched per stream, and a defined baseline of what everybody can do afterwards.
A short survey establishing real starting points
Groups streamed by ability rather than job title
Content pitched appropriately per stream
A defined baseline of organisational capability
Materials people can return to afterwards
Reporting on where ability still varies
Features
- Streamed by ability, not grade. Seniority predicts AI familiarity poorly, and grouping by it recreates the problem inside every session.
- Starting points established beforehand. A two minute survey before the session is what makes streaming possible, and it costs nothing.
- Smaller groups than convenient. Large sessions favour the confident. The people who most need training are the ones who go quiet in a room of forty.
- A stated baseline afterwards. What everybody can now reliably do is defined, so the organisation knows what it has rather than who attended.
- Materials that survive the session. A short reference people return to, since a full day of content is not retained from listening.
Benefits
- Everybody gains something. Streaming means the experienced are not bored and the beginners are not lost.
- A real baseline exists. You can state what your organisation can do rather than how many people attended.
- Quiet staff actually learn. Smaller groups reach the people who say nothing in a large room, who are usually the ones who needed it.
- Support requests drop. A consistent baseline reduces the informal asking that otherwise falls on a few colleagues.
Technologies we use
- Starting point assessment. A short survey establishing real familiarity rather than assumed familiarity.
- Streaming. Grouping by ability so content can be pitched rather than compromised.
- Applied practice. Working on real tasks within each stream rather than demonstrating features.
- Baseline definition. Stating what everybody should be able to do afterwards, and checking it.
- Reference materials. Short practical guides people return to when the session has faded.
The variable that decides whether organisation wide training works is group composition. Content matters less than whether the people in the room are at a similar starting point.
How the process works, step by step
Organisation wide rollout usually runs across two to six weeks depending on headcount, in sessions of half a day per stream.
-
Step 1
Free rollout planning
We establish headcount, tools licensed and how varied the starting points actually are.
-
Step 2
Survey before scheduling
A short survey establishes real familiarity, which is what makes streaming possible.
-
Step 3
Streams scheduled
Groups are formed by ability rather than department, deliberately smaller than would be convenient.
-
Step 4
Sessions delivered
Each stream works on real tasks at an appropriate level rather than a shared compromise.
-
Step 5
Baseline confirmed
What the organisation can now do is stated, and where ability still varies is reported.
Quality assurance and how we work
The commitments attached to training an entire organisation.
- We stream by ability rather than grade. Seniority predicts AI familiarity poorly, and grouping by it puts a director who has never opened a tool beside one who uses it hourly.
- We keep groups smaller than convenient. Large sessions favour the confident, and the people who most need training are the ones who go quiet in a room of forty.
- We survey before scheduling. Assuming starting points produces the mixed room that makes organisation wide training fail.
- We report where ability still varies. A completion figure tells you attendance. The gaps that remain are the useful output.
- We do not deliver a single session for everybody. It is cheaper and it is the format most likely to produce good feedback and no change.
Frequently asked questions about employee training
Why not run one session for everybody?
Because content pitched at the middle bores the experienced and loses the beginners. It is cheaper and it produces good feedback and no change.
How do you decide the groups?
By actual ability, established through a short survey beforehand. Job title predicts AI familiarity poorly.
Would grouping by department work?
Not well. Departments are as internally varied as the whole organisation on this particular skill.
How large should a group be?
Smaller than is convenient. In a room of forty the confident dominate and the people who needed it most say nothing.
How long does a full rollout take?
Two to six weeks depending on headcount, in half day sessions per stream.
What do we have at the end?
A stated baseline of what everybody can reliably do, plus reporting on where ability still varies.
Is attendance a useful measure?
No. It tells you who came. The remaining gaps are the output worth having.
What about staff who already use AI daily?
They go in a different stream and get content that is actually useful to them, rather than sitting through basics.
Will people be honest in the survey?
Mostly, if it is short and clearly not an assessment. We keep it to a couple of questions for that reason.
Can some streams run in Swahili?
Yes, and mixed delivery is common, particularly for operational rather than office based streams.
Do you provide materials afterwards?
Yes, deliberately short. A long reference does not get opened, and a full day of content is not retained from listening.
Can it cover our specific tools?
Yes, and it should. Teaching a tool you have not licensed wastes the session.
Does this include the security rules?
A short version, though organisations handling client data usually want the security session separately and in more depth.
What if managers do not attend?
It undermines the rollout. Teams follow behaviour, and a manager who did not attend signals it is optional.
Can we train trainers instead?
Yes, and it works well at large headcounts, provided the internal trainers get more depth than the streams they will teach.
What if people forget it all?
Likely without practice. The materials and a defined baseline help, and adoption work is what makes it stick beyond that.
Do you assess individuals?
No, and we would advise against it. Assessment turns a training session into an exam and people stop asking the useful questions.
Can it run across several branches?
Yes, and remote delivery works acceptably for this, though the smaller group rule still applies.
Is what staff say reported to management?
Patterns are, without naming individuals. A session where people fear being quoted produces no useful questions.
What does organisation wide training cost?
Headcount and stream count determine it, quoted after the rollout planning.
Glossary of terms
Rollout terms in plain language.
- Streaming
- Grouping learners by ability so content can be pitched rather than compromised.
- Starting point
- Somebody actual familiarity with a tool, which job title predicts poorly.
- Baseline
- What everybody in the organisation can reliably do after training.
- Applied practice
- Working on real tasks rather than watching features demonstrated.
- Rollout
- Delivering training across an organisation rather than to one team.
- Reference material
- The short guide people return to once the session has faded.
Related services in AI Training
Talk to us about employee training
Ask ten staff across your organisation how often they use an AI tool. The spread in the answers is what this has to handle.
Start your project