AI training for employees, built around their own work

Good AI training for employees covers how the tools work and fail, how to get reliable output on real tasks, what data stays out and how to spot work that a system should take over. Generic tool tours cover these badly, because people forget features they never try on their own work.

Our sessions are built from your material. Before the session we collect 15 to 30 real tasks from attendees, such as summarizing a client call, drafting a reply to a complaint, reconciling 2 spreadsheets or turning meeting notes into a proposal. Each exercise uses an anonymized version of one of those tasks.

A standard half-day runs in this order:

1. How language models produce answers, shown live on your documents, including where they make things up. Our entry on AI hallucination has the short version.
2. Prompt patterns that hold up: giving context, showing an example, asking for a format and asking the model to check its own work.
3. Practice on your tasks, in your approved tool, with us walking the room.
4. Data rules: what never goes into an AI tool, how to spot a risky request and what prompt injection looks like in a shared document.
5. Spotting automation candidates. Each attendee names 1 task they would hand to a system, and the list goes to their manager.

Full days add role tracks for sales, finance, support or operations, and a longer build session where teams create templates they will reuse.

Half a day is enough for 1 team that shares tools and tasks. A full day suits mixed groups, or teams that want to leave with finished templates. Anything under 3 hours turns into a demo, because hands-on practice is the part that changes habits.

Half-days run well over video. For a full day we suggest on-site, or 2 remote half-days a week apart, since 6 hours on a call wears anyone out.

We cap groups at 25 so every attendee gets help during practice. Larger teams run as repeated sessions with the same exercises, and because the preparation is shared, later sessions sit toward the bottom of the band.

The leadership session is shorter and less hands-on on purpose. Executives need enough practice to stop being impressed by demos, then time to decide the policy and budget questions that only they can settle.

Our corporate AI training costs $3,000 to $6,000 for a half-day and $6,000 to $12,000 for a full day, for up to 25 people. At a full room that is $120 to $480 per person, with exercises built from your own documents.

What moves the price within each band:

• Custom exercises. Exercises from your anonymized documents take more preparation than our standard set, and they are what makes the training stick.
• Role tracks. 1 shared track costs less than separate tracks for sales, finance and support.
• Tool setup. If your approved tool needs shared prompt libraries or custom assistants set up before the session, that adds preparation.
• Policy depth. The 1-page usage policy is included. A full acceptable use policy, checked against your existing IT and data policies, costs more.
• On-site delivery. Travel is billed at cost.

What is not included: licenses for AI tools, which you buy directly from the vendor; helpdesk support after the 30-day clinic; and certification. We do not issue certificates that claim to qualify anyone in AI. If you want a policy to start from before the session, our AI acceptable use policy template is free to use.

Measure what people do with AI on real tasks in the month after the session. Satisfaction scores on the day tell you whether people enjoyed it, which is a different question. We track 3 measures.

• Usage. Before the session we take a baseline from your AI tool's admin usage reports, where your plan includes them, and compare it 30 days later. Microsoft 365 Copilot, for example, has a Copilot usage report in the Microsoft 365 admin center (opens in a new tab) that shows enabled and active users.
• Task time. Each attendee names 1 task before the session and estimates how long it takes, then reports again at the 30-day clinic. It is self-reported, so treat it as a direction and look for the same task coming up across several people.
• Automation candidates. How many tasks people flag and how many a manager would act on.

We will not quote a productivity percentage after a single workshop, because a half-day of training cannot support one. If you want hard numbers, pick 2 or 3 tasks with countable volume, such as tickets answered or proposals drafted, and track them for a quarter.

The 30-day clinic is where problems surface: prompts that worked in the room but fail on messy real inputs, or a team that stopped because IT blocked a feature. We fix what we can on the call and send the sponsor a short note on the rest.

A workshop is the wrong choice if your staff do not yet have an approved AI tool. Training people on a tool they cannot use at work pushes them toward personal accounts. Sort out licenses and a basic policy first; we can help with both before booking the session.

It is also the wrong fit when:

• The real goal is to automate a process. Training makes individuals faster at a task. If the task is high-volume and follows rules, such as keying invoices or routing leads, build an automation and train the few people who supervise it. See AI automation.
• You want a keynote for 200 people. We cap sessions at 25 so everyone gets hands-on time. For large groups, a short talk followed by repeated team sessions works better than 1 big room.
• You need technical training for developers, such as building with model APIs, retrieval or evaluation. Our workshops are for people who use AI tools at work. Developer skills grow faster inside a real build, where your engineers pair with ours on custom AI agents or a knowledge system.

What is not included: AI tool licenses, IT configuration beyond preparing shared prompt libraries, and ongoing training after the 30-day clinic. If tools change and you want a refresher, book a repeat session at the same rates.