Harvard/BCG Study: AI Training Gave Juniors +43% — How to Replicate the Design

When a founder of a 40-person ops team told me they’d trained juniors on AI but saw no change in output, I realized they’d skipped the structured practice loop that makes training stick.

How to structure the first AI training week for juniors

Start with a 90-minute session focused on prompt basics and three concrete use cases from their daily work. Have each junior pick one repetitive task they do at least twice a week — like compiling weekly sales reports, filtering job applications, or drafting standard client emails.

In the second session, run a Shoulder-to-Shoulder hot seat for three volunteers. The group observes as the employee shares their screen, explains the current manual steps, and then experiments with AI to replace the repetitive parts. For example, a junior in HR might use AI to draft the first version of a job description based on a template, then spend 5 minutes editing it instead of starting from blank.

Tool tip (Course for Business):

In our corporate AI intensive, we use the Augment, don't replace framework to ensure employees see AI as a tool for their expertise, not a threat to it. During the Shoulder-to-Shoulder hot seat, participants immediately apply AI to a real task and leave with a working automation. See how the first week maps to your team’s first three tasks.

How to measure what actually changed

After week one, ask each junior to track time spent on their chosen task for three days before training and three days after. Calculate the average time per instance. The goal is not to eliminate the task but to reduce the repetitive portion — for example, cutting report compilation from 40 minutes to 22 minutes by using AI to pull data and format tables.

Avoid measuring “AI usage” or “prompt count.” Instead, measure time saved on the specific workflow they improved. If a junior previously spent 60 minutes per week on a task and now spends 35, that’s a 42% time reduction — the kind of gain seen in the Harvard/BCG study when training included hands-on application to real work.

Manager scan (what AI champions report after week 1)

• Champions observed juniors spending less time on repetitive steps like data gathering and formatting.
• Juniors reported higher confidence in starting tasks because AI handled the blank-page phase.
• Managers noted fewer requests for help on routine tasks, freeing up their own time.
• Time saved per junior ranged from 20 to 50 minutes per week on their selected workflow.
• Champions spent 60-90 seconds per day answering quick AI questions from peers.
• No juniors reported using AI for decision-making or customer-facing communication without review.

Micro-case (what changes after 7–14 days)

A 45-person professional services firm ran the first two weeks of AI training with juniors in operations and HR. After the Shoulder-to-Shoulder sessions, juniors began using AI to draft first versions of internal reports and job descriptions. Managers noticed that drafts arrived 20 minutes earlier in the day, giving them more time to review. Juniors said they felt less resistance to starting repetitive tasks because the initial effort was lower. After two weeks, the firm had three working automations built by juniors themselves — each tied to a specific, repeatable task in their workflow.

Note on this case: This example is illustrative — based on typical patterns we observe with companies of 30–500 employees, not a single named client. Specific numbers are rounded approximations of common ranges, not guarantees.

FAQ

What if juniors say they don’t have repetitive tasks? Every role has repetitive elements — look for tasks done the same way weekly, like updating trackers, formatting responses, or compiling lists. Start there.

Should we train seniors first? No. Juniors benefit most from early AI training because they perform more routine work. Seniors can join later to apply AI to complex workflows like forecasting or strategy.

How much time should we allocate for training? A minimum of 4 hours spread over two weeks — split into two 90-minute workshops plus optional hot seats. More time helps, but the key is applying AI to real work during the session.

What if juniors use AI incorrectly and make mistakes? Mistakes are part of learning. Review outputs together, focus on improving the prompt or adding a verification step, and treat errors as data — not failure.

Conclusion

Replicating the Harvard/BCG design means giving juniors structured time to apply AI to their actual work — not just theory. Start small, measure time saved on specific tasks, and let employees own the automation they build.

If you want your team to finish with working automations they built themselves — book a 30-min call and we'll map your first three tasks.