Somewhere under the sky, a course that AI built in two days is waiting on a four-minute voice note. The voice note is about “slide 3, the one with the chart.” There are four charts.
Down the hall, someone who just finished their company’s AI training is staring at a spreadsheet, wondering what exactly they’re supposed to do next.
Both of them are living with the same problem. They just don’t know it yet.
The good news first
Let’s give AI its due. Course production is faster than ever. SMEs and business teams can turn an idea into an interactive course in a fraction of the time it used to take. And most organizations have rolled out AI fluency programs, so their people know what prompting is and how a large language model works.
AI did exactly what it promised. It made the fast part faster.
Gap one: the course that builds itself
Here’s the catch. Faster production didn’t remove the bottleneck. It moved it.
When a course comes together in days, two questions get louder:
- Who makes sure it actually teaches?
- Why does review still run on screenshots with red circles, and email threads titled “Re: Re: Re: Feedback_FINAL_v7”?
We’re hearing the same pattern from L&D teams over and over again: the build takes days, and the review takes weeks.
We automated the drafting. We forgot to upgrade everything that happens after “Send for review.”
Gap two: the team that knows AI but can’t use it at work
Meanwhile, AI fluency is done, and a new question has arrived in every function: “Great. How do I use this in my actual job?”
The answer depends entirely on the job:
- Finance wants automation and dashboards.
- HR and Legal want AI-assisted review and responsible use built into their workflows.
- Tech teams want agents, agent loops, and RAG.
- Product and customer service want use cases built around the problems they solve every day.
Generic training explains the tool. It doesn’t explain the job.
Knowing what an LLM is and knowing what it can do for month-end close are two very different skills.
The common thread
Put the two gaps side by side and they look remarkably similar. In both cases, AI handled speed. What’s missing is expertise in the loop:
- an instructional designer who owns the learning,
- an SME who reviews the real thing,
- a practitioner who knows how the work gets done.
Speed is now a commodity. Judgment isn’t.
What good looks like
For building courses:
- AI-assisted production, with instructional designers owning the strategy, structure, and design decisions.
- SMEs review the actual course screen by screen, commenting in context instead of in scattered documents.
- The output stays highly interactive and SCORM-ready, so it works in any LMS.
For building capability:
A blended rhythm shaped around each function.
- Learn the foundations through short self-paced modules.
- Explore with an expert in a focused session on real workflows.
- Apply to relevant scenarios.
- Reinforce it with resources that bring the learning back into work.
At Apposite, this is the part we care about most: not replacing expertise with AI but putting expertise exactly where it is needed.
Three questions to ask your team this week
- Who owns instructional quality in our AI-built courses?
- How does SME feedback actually reach our designers?
- What does each function need to learn after AI fluency?
If the answers are “not sure,” “voice notes” and “the same thing as everyone else,” you’re in good company. That’s where a lot of teams are right now.
Back to slide 3
Picture the same two people a few months from now. The SME leaves a comment directly on the chart that needs fixing, and the designer knows exactly which one. The person at the spreadsheet has built their first automation, alongside someone who’s done it before.
AI did the fast part. The rest is still up to us, and honestly, that’s the most interesting part.
Want to see what that could look like for your team? Get in touch with Apposite, and let’s start with the part AI didn’t solve.
.