For years, the job of a Learning and Development team was to make things: slide decks, PDF guides, LMS modules. You wrote content, packaged it, uploaded it, and hoped people clicked through it. The output was always static. The feedback loop was slow.
That’s changing, and the change is more specific than “AI is transforming L&D.”
Tools like Claude Code, Claude Sonnet, and OpenAI Codex do more than speed up content production.
They essentially turn L&D teams from static content creators into builders of interactive learning experiences and tools. You can build software without needing a developer. That’s the actual shift. You can now produce a working interactive app faster than you could previously produce a PowerPoint deck. In our case, we built 50 interactive online exercises for our “Writing policy and procedures” elearning course in under 90 minutes, while sat on the sofa.
What these tools are, briefly
Claude Code is Anthropic’s command-line tool for agentic coding. You describe what you want, it writes and iterates on code across files. OpenAI Codex works similarly. It turns natural language into working software. They are different from chatbots with a code mode. They’re tools that can build an entire application from a prompt.
For L&D professionals, that distinction matters. You’re not restricted to asking an AI to write you a quiz. You can ask it to build you a quiz application, complete with branching logic, scoring, and a results screen.
Use Cases worth your time
Interactive courseware that actually works
A cybersecurity awareness course built as a web application is a different product from a PDF about phishing. Learners click on a simulated phishing email. They make a decision. They get immediate feedback based on what they chose. The experience has consequences, even if they’re fictional.
These tools can build that. You describe the scenario structure, the decision points, the feedback messages. It produces the code. You test it, refine it, and ship it.
The same applies to compliance training, product knowledge courses, and technical procedures. Any course that benefits from “what would you do?” moments is a candidate.
Onboarding portals that adapt by role
A new sales hire and a new engineer have almost nothing in common in their first two weeks. They need different tools, different documentation, different introductions to the business. A single onboarding deck serves neither well.
With these tools, you can build a lightweight onboarding portal that surfaces different content depending on role. The engineering required is modest. A few conditional displays, a simple intake form, a curated set of resources per path. The tools can scaffold this in an afternoon.
Training dashboards that show what’s actually happening
Most L&D teams report completion rates because that’s what their LMS gives them. Completion tells you almost nothing about whether anyone learned anything or whether the training was worth the time.
A custom dashboard can pull data from your LMS, your HRIS, and your performance management system, and surface something more useful: skill gap analysis by team, course drop-off points, correlation between training completion and performance metrics.
Building this used to require a data engineer and a front-end developer. Now, an L&D analyst who understands their data can produce a working dashboard by describing what they need.
Simulations for skills that require practice
Sales coaching, difficult conversations, technical support calls are skills that don’t develop by reading about them. They develop through repetition and feedback. Learning by doing.
A branching simulation gives learners a scenario. They choose a response. The simulation reacts. This is not new pedagogically, but it has historically been expensive to build. Articulate Storyline files take weeks. You could reduce that to hours.
The quality ceiling is higher than it was, too. Instead of a clunky multiple-choice branch, you can build a simulation with free-text responses, AI-driven feedback, and conversation histories that a manager can review.
Just-in-time tools, not courses
Not everything needs to be a course. Sometimes someone needs to know how to raise a purchase order right now, while they’re trying to raise one.
A small, task-specific tool (for example: how to raise a ticket, how to submit expenses, what to do when a customer asks for a refund) is faster to build and more useful than a module buried in an LMS. These tools can produce these as standalone web pages, embeddable widgets, or Slack-integrated tools.
Where L&D teams need to shift
The technical barrier is lower. The design and instructional thinking still matters as much as it ever did.
Building an interactive simulation badly is worse than not building one at all. A branching scenario with poor feedback, confusing logic, or irrelevant scenarios teaches people nothing and wastes their time. The AI writes the code. The L&D professional still needs to design the experience.
The new role looks more like product ownership than content creation. You decide what the learner needs, what the flow should be, what feedback is useful, and what success looks like. The AI handles implementation. That’s a different skill set from scripting a video or building slides.
It also means L&D teams need to get comfortable testing and iterating on software products. That means user testing with a sample of learners before rollout, tracking usage data, and updating the tool when the workflow it supports changes.
A practical starting point
Firstly, training that improves your skills can help – AI for learning design: a hands-on course for in-house L&D teams.
If you want to test this without committing to a full project, start with something small and self-contained.
- Pick a process that people frequently ask about. For example: expenses, IT requests, booking meeting rooms. Something that generates the most “how do I…” questions in your organisation.
- Write out the steps clearly. Describe what you want: a simple web tool that walks someone through the process, checks their understanding at each step, and gives them a checklist at the end.
- See what it produces.
- Iterate on it. Show it to two or three people who actually do that task.
That’s the loop. It’s faster than you think, and the output is more useful than most of what a traditional authoring tool produces.
The shift for L&D is real, but it’s not magic. These tools remove the need for a developer. They don’t remove the need for someone who understands learning, knows the audience, and can tell the difference between a simulation that teaches and one that merely entertains.
That person is you. The tools just mean you can finally build what you’ve always wanted to build.
Leave a Reply