AI training for employees, built on your team's own work

Who is AI training for employees for?

AI training for employees is for a business whose people have AI tools, or are about to, and use them unevenly. Some staff paste work into ChatGPT every day while others have never opened it, and nobody has shown the team what the tools do well on its own work, what they get wrong and what should never be pasted in. Training closes that gap, and it takes a different shape in each of three kinds of business:

• Established local businesses: one team, often most of the company, in one workshop built on the jobs it does every week.
• Multi-location and regional businesses: a few teams doing the same work in different places, trained over several sessions so each site can try the same methods and compare what works.
• National businesses: many departments, a leadership team and rules about data, where training runs across the company with a briefing for leaders, sessions for each department and a group of champions who keep it going.

If your business hasn’t settled on any AI tools yet, the free AI readiness worksheet is a better first step. It takes one task through five parts (the work, the data, the systems, the people and the economics) and scores how ready it is for AI.

What does AI training for employees cover?

The training is about the work, not the software. We start from the tasks that take your team’s time, such as drafting replies, summarizing long documents, pulling figures out of reports, preparing quotes or writing up meetings. Then we show what the AI tools you already have can do with each one, and where they fall short.

From there, the sessions cover the habits that make the tools useful at work: giving a tool the context a task needs, checking what comes back against the source, knowing which tasks to keep away from AI, and knowing what your business allows to be pasted in. People practice on their own examples and keep the prompts that worked. By the end, each person should know which of their tasks AI helps with and which it doesn’t.

Is AI training about people or about training a model?

People. In computing, “AI training” also means training a model on data, which is a different kind of project and not one we take on: we don’t train AI models. This service teaches your employees to use AI tools that already exist, on the work they already do.

If what you need is software that does a task by itself, that’s AI automation built into your systems. Training can come before it, so the people who will work beside an automation know what it does and what to check.

What happens in an AI workshop for one team?

A workshop is the market’s entry format: one team in one live session, on site or by video. At 2020solutions.ai and NEU Media Group it runs a half day or a full day, and build.ms also sells sessions of an hour or 90 minutes. NEU Media Group builds every session on the team’s real tasks.

A workshop for your team follows the same shape. Before the day, we go through the work with you and pick the tasks the session will use. On the day, people work through those tasks with their tools open, and leave with the prompts and notes made for them. A follow-up afterwards checks which habits stuck and what stood in the way.

How does a multi-session AI training program work?

A workshop gives a team a start. A program gives it time to change how the work gets done. The market sells this as an adoption program, a rollout or a cohort. 2020solutions.ai sells adoption as a multi-week engagement, and Albright Labs sells its founding cohort as six weekly live sessions of 60 minutes for 10 to 15 people. Justin McKelvey’s rollout is a different shape: for teams on ChatGPT Business or Copilot, he sets the tools up himself, then trains the team live, for 60 minutes, on the workflows he built.

A program for your teams runs alongside the rollout of the tools you’re paying for. Each session takes one part of the work, such as the shared inbox, call notes or the monthly report, and changes how it’s done, with check-ins between sessions on what is working and what isn’t. It suits a few teams doing similar work, or one team that needs more than a day to change its habits.

What does company-wide AI enablement involve?

Enablement is the market’s word for training across a whole company, the third size corporate AI training is sold in, after the workshop and the program. 2020solutions.ai, for one, lists an enablement program of several sessions among its corporate offers. Enablement reaches every department, so it needs more than sessions: leaders have to agree on what the tools are for, and someone inside each team has to keep things going once the trainers leave. A company-wide program has three parts:

• a short executive briefing for the leadership team;
• sessions for each department, on its own work;
• a group of internal champions, people in each team who carry the new ways of working on and answer their colleagues’ questions.

It takes more preparation than the other two levels, because each department’s work is different and each set of sessions is built for it.

What should AI training for executives cover?

Leaders don’t need the same session as the people doing the work. An executive briefing is shorter and covers the decisions only leaders can make: which tasks AI should and shouldn’t touch, what the tools will cost across the company, what data stays out of them and who answers for the results. It also shows leaders what their teams are being asked to change, so they can back it.

If your leadership team wants a written plan for where AI goes next, with costs and an order of work, that’s a separate piece of work: an AI strategy and roadmap. We don’t give governance or compliance sign-off.

Where does data and AI upskilling fit?

Upskilling is the word for teaching people new skills at work, and data and AI upskilling is sold in two very different forms. One is a library of self-paced courses: Coursera and Udemy sell yearly licences to their course libraries, priced per person. The other is live training on a team’s own work, which is this service. A library suits someone who wants to study a subject at their own pace. Live training suits a team that needs to change how a particular job gets done, with the tools and data it already has.

On the data side, the skills that matter at work are plain ones: knowing what a tool needs from your records to give a useful answer, checking an answer against its source, and knowing which data shouldn’t leave your systems.

How do you measure whether AI training worked?

By the work, not by attendance. Before the training, pick a few tasks it will cover and write down how long each takes and how often it needs fixing afterwards. At the follow-up, look at the same tasks again and ask who is still using what they learned. That gives you an answer for your own business, which a general figure can’t.

Set it against what the training costs in full: the sessions themselves, the seats for the tools, which your business pays the software company for, and the hours your team spends in training. If the numbers don’t add up for a task, that’s worth knowing too, and the follow-up is the place to say so.

What isn’t included in AI training?

The training covers your people. Most of the market leaves these out, and so do we:

• Tool seats and licences. Your business buys ChatGPT, Copilot or Gemini seats directly from the company that makes them.
• Building automations. Software that does a task by itself is done-for-you AI work, scoped and priced on its own.
• Certification. There’s no exam to pass and no certificate at the end.
• A course library. The sessions are live and built for your team, not a set of recorded lessons.
• An AI policy. Writing one isn’t part of training.

Where do we use AI ourselves?

In AlvyOne, the operations system we built and run with Alvy. Both Alvy Logistics and DSI Transportation work in it. Alvy is owned by our co-founder Noah Alvarez, and DSI is an Alvy company. The dispatch team there works with an AI assistant in its team chat every day: it finds jobs, answers status questions, and edits, cancels and assigns drivers, with a confirm step before a risky change. That’s software we built, not a training engagement. It’s also why the training puts weight on checking what a tool did before anything changes, since the assistant asks for that confirm for the same reason. The AlvyOne case study covers the system it works in.

When is AI training the wrong call?

Training is the wrong first step in a few cases, and if yours is one of them, we’ll say so on the call:

• Nobody has decided which tools are allowed. Training people on tools they can’t use at work wastes the session. Settle that first.
• The task should run without a person typing it. If a job repeats all day and follows clear rules, building it into your systems may be worth more than teaching people to do it faster.
• The process is the problem. AI won’t fix work nobody agrees on how to do. An AI readiness assessment maps the work first and ranks where AI fits.
• You need one use of AI built and running. That’s AI implementation, where the work is planned, built and run rather than taught.

How does AI training for employees start?

With a 15-minute call. You don’t need a training plan before it. If you’re weighing training against other AI work, our page on what AI consulting costs gives the market’s ranges for that work. Book a 15-minute call to talk through your team.