Building AI fluency and adaptability at an SMB

Straight to the Story

In the second quarter, Tom announced that the company would adopt AI tools in support, sales, and finance. Maya’s first reaction was concern. Without a plan, some people would experiment on their own, some would avoid the tools, and nobody would agree on what was safe to use. The product launch was also six weeks away, and support was already stretched.
‍

She had reason to treat this as urgent. In US job postings, demand for AI fluency has risen about 11 times since 2022. Demand fora daptability is up about fivefold, and demand for resilience, curiosity, and willingness to learn is up about threefold. Workforce researchers group these under the name empowering skills, because they prepare people for continuous change rather than one specific job.
‍

The research gives more direction here than for the other two skill types. It says empowering skills can be intentionally developed. For AI fluency in particular, it points to organizations that have moved past ad hoc training toward structured learning pathways and role-specific skills. For the other four, most of the evidence comes from studies of children and teenagers, so the practices below for adults at work are our recommendations.

‍

The Next Steps
‍

AI fluency

1. Start with one recurring task per person. Maya invited a first cohort of ten, including Priya. Each person picked a task they did every week, tried an AI tool on it, and wrote down what they asked, what came back, and what they changed. Practice on real work in your own role is what structured programs aim for, and it beats a generic course.

2. Hold a monthly demo. Two people showed the group what worked and what did not, in thirty minutes. The failures taught as much as the wins.

3. Write one page of guidance. Tom had security concerns, so Maya and the head of IT wrote a one-page note on which data could go into which tools and who to ask. People experiment more when the rules are clear.

4. Add a structured baseline if you want one. The large technology providers, including Google, IBM, and Oracle, publish AI learning paths.

‍

Adaptability

Maya created low-stakes exposure to unfamiliar work. One specialist covered a sales operations teammate’s tasks for a week. Each team picked one process to change in the quarter and designed the new version themselves. People adapt faster when they have done it recently and the stakes were small.
‍

Resilience

The launch arrived on schedule and support volume jumped. In the monthly workload check, scores in Luis’s team dropped sharply. Maya and Luis did not push on. They removed a weekly report that nobody read closely, moved two cohort demos back by a month, and told the team that learning hours still counted as work. People cannot build new capabilities on a foundation of burnout or overload, and managing that load is the employer’s job. Check workload in every one-on-one, run reviews after incidents without blame, and when you add a learning expectation, remove something.
‍

Curiosity and willingness to learn

Maya started a weekly “what I tried” thread in the team chat. Tom posted first, with an AI-written summary that had gotten a customer’s name wrong. After that, others posted their own misses. Four hours a week of learning stayed on the calendar for the cohort, and each participant’s manager protected it.

Three things to track: how many tasks each participant has changed, how many hours that saved and where they went, and an employee-reported workload score each month. If workload goes up while hours saved go down, adjust the plan before the next cohort starts.

‍

In Part 4, two quarters have passed and Tom asks whether any of this is working.

‍