Skip to main content
Two men plan in front of a wall of yellow and blue sticky notes, one arranging notes while the other thinks
All postsTraining

AI workshop for a startup team: session plan

Who should attend a startup team’s first AI workshop, and what to prepare. How a three hour session can run, and the check-in after.

What should a first AI workshop for a startup team cover?

A first AI workshop for a startup team should cover three things. What the approved tools do well and where they guess. The one page rule on which data stays out of them. Then two tasks from each person’s own week, rebuilt with the tool. Everyone should finish at least one of those tasks before the session ends.

A startup team of 5 to 50 people has little time for training. It has even less patience for a session that changes nothing. A workshop earns its afternoon when Monday’s work looks different. That happens when the exercises come from the team’s own tasks and files.

This post is the session plan: who attends, what to prepare, how the time runs, and what comes after. How long training takes and what the written policy covers are in AI training for employees.

Who should be in the room for a startup AI workshop?

Invite the people who do the most repeated writing and admin. That means the person who drafts proposals, the one on support email, and whoever runs operations or finance. Add a founder or manager who can settle a tool or data question on the spot. A group of six to twelve lets everyone get help on their own task.

In a team under fifteen, that can be everyone. In a team of forty, run the first session with one group and a second session a few weeks later.

The founder matters more than the headcount. In a startup the founder often picked the tools and holds the customer contracts. When someone asks whether the contract can go into the assistant, the session stalls unless a person in the room can answer. A founder answers it once, for everyone.

Anyone who already uses AI tools every day belongs in the room too. Seat them as a helper, so the person who runs the session is not the only one who can help.

Some teams work mostly in office roles, such as admin, finance or customer service. For them, AI training for office staff covers the tasks those roles bring.

What should the team prepare before an AI workshop?

Prepare three things a week ahead. Pick the approved tools and set up a business account for each person. Write a one page rule on which data stays out of those tools. Ask each person to bring two tasks from their own week, with the files they use. Without the tasks, the session runs on made up examples.

Approved tools, with accounts ready

Decide which assistant the team will use. If the business already pays for an office suite, check which assistant comes with the plan before buying another one. Set up every login before the day. Twenty minutes of password resets uses up a ninth of a three hour session.

Naming the tools matters because people already use AI at work, approved or not. WalkMe surveyed 1,000 US workers who use AI in their jobs in July 2025. Of them, 78% said they use AI tools their employer did not provide. A workshop on approved tools gives that use accounts the business controls.

Source: WalkMe, Employees Left Behind in Workplace AI Boom, New WalkMe Survey Finds, 2025.

A one page data rule

Keep it to one page and read it out first. It names the data that never goes into any tool, the approved tools, and the person to ask when a case is unclear. The policy section of the employee training guide lists what the page should cover.

Two tasks per person

Ask each person for two tasks they do every week that involve writing, summarizing or moving information. Ask for the files too: last week’s proposal, a support thread, the notes from a customer call. Each task should be small enough to finish in half an hour.

Good startup examples:

  • A follow-up email after a sales demo
  • Release notes from a list of tickets
  • A summary of customer interview notes
  • The first draft of a monthly investor update
  • A reply to a support question asked every week

The same WalkMe survey has a second finding. Nearly 60% of respondents said it often takes longer to figure out how to use AI than to do the task by hand. The session is built around that problem: each person works on their own task, with help nearby.

Source: WalkMe, Employees Left Behind in Workplace AI Boom, New WalkMe Survey Finds, 2025.

What does the AI workshop session look like?

Most of the session is hands-on. A short opening covers the data rule and what the tool does well. Then each person works through their two tasks on their own files, with help nearby. They check a partner’s output for errors. The session ends with each person writing down the steps that worked, so the team can reuse them.

Here is an illustrative agenda, not a fixed program. It fits a three hour afternoon with eight to twelve people:

  1. 15 minutes. The data rule, read out, with questions.
  2. 30 minutes. What the tool does well, shown live.
  3. 15 minutes. Log in, attach a file, save a prompt.
  4. 45 minutes. First task, each person on their own files.
  5. 10 minutes. A break.
  6. 40 minutes. Second task, then partners check each other’s work.
  7. 25 minutes. Write down the steps that worked.

The two task blocks take 85 of the 180 minutes. With the log-in and the write-up, people spend about two of the three hours working with the tool, not watching. The demonstration uses a task someone brought, so the group watches its own work, not a stock example. The pair check builds the habit that protects customers: reading every draft before it goes out. In the last block, the team also agrees where saved prompts and written steps will live.

What each person leaves with

  • Two tasks done with the tool, on their own files
  • The written steps for each task, saved for the team
  • A working login to the approved tool
  • The one page data rule
  • One name to ask when something is unclear

What follow-up should come after a startup AI workshop?

Book a one hour check-in one to two weeks after the workshop, while the tasks are fresh. Each person shows the last time they used the tool on one of their two tasks, and the group fixes what failed. Each person then adds one new task. Without the check-in, nobody knows which tasks stuck.

The first check-in

Ask each person to show the work, not describe it. A task nobody has repeated was the wrong pick, or the steps were unclear. Rewrite the steps together, or swap the task for one that comes up more often.

A month later

Before the workshop, ask each person to note roughly how long one of their tasks takes. A month later, ask again. How to measure AI ROI in a small business shows how to count it without guessing.

When a prompt is not enough

Some tasks keep coming back as the same question: the pricing rules, the onboarding steps, the product specs. That points to an assistant that reads your own documents, which how to train an AI assistant on your business walks through. Other tasks turn out to be a process to automate, not a prompt to improve. Which processes to automate first helps sort them.

Tool workshops and a written AI use policy are part of the AI training on our services page.

Is there an AI workshop for a startup team in Orange County?

Yes. Preferred AI Partners is based in Irvine and works with businesses across Orange County, training included. Meetings are online or in person, depending on the situation. A first workshop is built on your team’s own tasks. The Orange County page has the details.

The plan above holds wherever the session runs. Pick the tools, write the data rule and collect two tasks per person before the workshop starts.

More on this

Share this post

Sebastian Alidad

Founder, Preferred AI Partners

Sebastian is a founder of Preferred AI Partners and runs the AI audit call.