Free tools Windows power users keep installed
One-click scans. No signup required.
AI upskilling fails when employers treat it as a training assignment rather than a change in how people do their work. Employees need to know what problem a tool solves, have protected time and safe resources to practice, and get support as they apply what they learn. In a September 2025 episode of What IT Leaders Want, Redgate Software’s Jeff Foster describes practical ways his organization tries to make that possible.
Why one-off AI training falls short
A video or presentation can introduce a tool, but it cannot by itself teach someone when to use it, how to judge its output, or what to do when a situation does not match the lesson. Foster argues that people learn technology more effectively when they first understand the problem it is meant to solve. As he puts it, “you need to understand the why of that technology.”
As an Amazon Associate I earn from qualifying purchases.
He uses Kubernetes as an example: without a grasp of the problem container orchestration addresses, Kubernetes can look like an unnecessarily complicated solution. The same principle applies to AI training. Start with a work problem employees recognize, then show where AI might help, where it may not, and how to assess the result.
What the available figures say—and what they do not
Pluralsight’s March 2025 announcement reported findings from a survey of 600 technology decision-makers. Those responses suggest that skills concerns are a practical barrier for some organizations, but they are not a census of all employers or a measure of every worker’s ability.
#1 Best Overall
| Reported finding | What it describes |
|---|---|
| 75% of companies had experienced delays or pauses in at least one AI project because of a lack of employee AI expertise. | Pluralsight’s reported 2025 survey responses from 600 technology decision-makers. |
| 35% of respondents said half or fewer of their employees had well-developed AI skills. | Respondents’ assessment in the same Pluralsight survey. |
| 38% of respondents said half or fewer of their departments had adopted AI skills in training programs and day-to-day use. | Respondents’ assessment in the same Pluralsight survey. |
Read Pluralsight’s March 6, 2025 announcement. The episode also mentions Microsoft and LinkedIn findings, but the exact original source for those figures is not established here, so they are not repeated as verified statistics.
How Foster says Redgate makes learning part of work
Foster’s examples come from his account of practices at Redgate; they are not independently audited outcomes or a guarantee that the same approach will work in every organization.
Make time available
Foster describes “10% time,” with Friday afternoons reserved for learning and development. People might give lightning talks, try a technology, build a small application, or tackle a customer issue in a new way. The transferable lesson is to put learning on the calendar rather than expecting employees to fit it around a full workload.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Give people bounded chances to practice
Small, realistic experiments can help employees move beyond abstract instruction. Foster mentions a toy application and a Slack bot that orders team lunch. A leader can choose a low-risk task, define what data and tools are allowed, and make it clear who can help if an experiment goes wrong. This gives people room to learn while keeping the exercise within organizational guardrails.
Keep the reasoning behind decisions
Foster says Redgate asks people making software changes to document why they are making a change, which alternatives they considered, and why they chose one. These records are called architecture decision records, or ADRs. Foster said the library had almost 500 decisions when he spoke; that is a historical statement in the episode, not a current count.
For AI-related work, a similar record can capture the problem being addressed, the options considered, the reasons for a choice, and the relevant risks or constraints. This helps colleagues understand the context rather than copying a decision as a rule without knowing when it applies.
Rank #4
Hire and develop for curiosity
Foster says Redgate looks for people who show a thirst for learning. His interview examples include asking candidates about the last book they read, a technology they have tried, and what excites them. These are examples from his account, not a validated hiring test. For current employees, the broader point is to notice what people want to learn and create supported opportunities that connect those interests to work.
Build judgment, not just confidence
Learning a few rules can make a beginner feel certain before they have seen enough situations to know where those rules break down. Foster calls this the “expert beginner” problem: “they’re confusing confidence for competence.” He argues that people need supported exposure to varied situations, including cases where a familiar approach is insufficient.
His example involves software used in customer environments with very different database counts, including an unusually large deployment. The point for leaders is not to manufacture edge cases, but to let employees encounter meaningful variation with guidance, so they can understand why a decision that works in one setting may fail in another.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.A practical way to assess an AI learning plan
The episode does not rank training methods through comparative testing. Its examples do, however, offer useful questions for assessing whether a program is likely to support real skill-building:
- Work connection: Does the learning address a problem employees actually need to solve?
- Time and resources: Is there protected time, access to appropriate tools, and clear direction on what is safe to use?
- Practice and support: Can employees try the skill on a bounded task and get help when they are unsure?
- Judgment: Does the program teach people to evaluate situations and outcomes, rather than memorize rules?
- Shared learning: Can colleagues revisit the reasoning behind previous decisions and learn from it?
These are checks, not a formula for guaranteed results. The available episode transcript covers Foster’s career-goals discussion only in part, so it does not establish a fuller set of specific recommendations on that subject.
Why leaders need to provide safe direction
The episode’s hosts warn that employees may experiment on their own if an employer does not provide time, tools, resources, and guidance. Matt Egan raises the risk of staff entering company information into public AI tools. The episode does not quantify the likelihood of that happening or provide a technical security assessment; the practical implication is that organizations should make permitted tools and data-handling expectations clear while giving employees a supported way to learn.
The episode, hosted by Keith Shaw and Matt Egan, was published by CIO on September 12, 2025. Its central lesson is that upskilling is not simply a matter of assigning courses: leaders have to connect learning to work, make practice possible, and help people develop judgment over time. Watch or read the episode.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
