The Frontline L&D Team of 2028 Has Three People and 40,000 Learners. Here's What They Do All Day.

Our position: within two to three years, the frontline learning function at most large companies will be a very small team running a very large operation, because the two things that used to consume L&D headcount, making content and chasing completion, are being automated out of the job. That is not a threat to L&D. It is the first time the function has had a chance to do the work it was hired for. But only if it lets go of the work it was doing instead.

I run Leap10x, and I want to be careful here because "AI will shrink your team" is exactly what a software vendor would say. So let me be precise about the claim. I don't think L&D people become unnecessary. I think the ratio changes by an order of magnitude, and the job changes with it. I've watched it start to happen in our customers over the past eighteen months, and I'd rather describe it honestly than pretend the old operating model survives.

Where L&D time actually goes today

Ask a frontline L&D team of ten what they do, and the honest breakdown at most companies looks something like this.

• Content production: three to four people. Taking SOPs, product decks and policy PDFs from operations and legal, and turning them into slides, videos and quizzes. Then translating them. Then updating them when the SOP changes, which is constantly.
• Delivery logistics: two to three people. Scheduling classroom sessions, booking trainers, coordinating with plant and store managers to release people from shifts, running train-the-trainer cascades.
• Completion chasing and reporting: two people. Uploading rosters, assigning modules, sending reminders, exporting completion reports for audits, reconciling the LMS against the HR system.
• Actual learning design and performance analysis: whoever is left, in whatever time is left. Usually not much of either.

This is not a criticism of the people. It's a description of what the tools required. An LMS that needs a roster upload every Monday, a content pipeline that needs a designer for every module, and a delivery model that needs a room and a trainer will eat a team of ten without ever getting to the question of whether anyone learned anything.

What has changed, specifically

Three of the four buckets above are now largely automatable. Not "will be". Are.

Content production. Upload a PDF, a slide deck or a Word document and get back a sequence of short lessons with quizzes, polls and voice cards, translated into the languages your workforce speaks, in under fifteen minutes. That is what our platform does today, and we're not the only ones; AI authoring has become a standard feature across the category in the past two years, as our comparison matrix notes. The quality varies, and a human still needs to review the output. But the three-to-four-person production bucket collapses to a review step. We described the workflow in converting training PDFs and SOPs into WhatsApp microlearning.

Delivery logistics. If training arrives on the worker's own phone in the gaps between tasks, there is no room to book, no shift to release, no cascade to run. The two-to-three-person logistics bucket doesn't shrink; for routine knowledge it disappears. Hands-on skills still need a floor and a supervisor, and we're clear about that in when WhatsApp training is the wrong choice. But the share of frontline training that is knowledge rather than manual skill is much larger than most L&D teams assume.

Completion chasing. Automated nudges, real-time dashboards by region and person, exportable audit logs. The chasing was never a human job; it was a job humans did because the LMS couldn't. Our completion rates run above 85% against the industry's 20 to 30%, and almost none of that comes from anyone chasing anyone.

What's left is the fourth bucket. Which was always the job.

What the three people do

Here is what I think the frontline L&D team of 2028 looks like at a company with 40,000 frontline workers. Three roles, not necessarily three people, but three is a realistic headcount.

The performance analyst

This person's question is: what is the business losing because workers don't know something, and where? They live in the data: quiz scores by site, completion by manager, near-miss reports by shift, customer complaints by store, sales by product. They find the gaps. "Stores in the western region fail the returns-policy question at three times the rate of the north, and their refund disputes are up." That's a training brief nobody wrote a year ago because nobody had the data or the time.

This is the role we designed our analytics for, and it's the one we see emerging first. Our piece on measuring training impact through chat-based analytics describes the data this person works with.

The content owner

Not a producer. An owner. This person decides what the 40,000 people need to know, in what order, at what cadence. They take the analyst's brief and the operations team's SOP and turn it into a lesson in an afternoon, review the AI's draft for accuracy and tone, decide whether it needs a voice card for the low-literacy sites, and ship it. They maintain the library so that when the SOP changes, the lesson changes the same day. The ADDIE model compressed to a five-day sprint is roughly their weekly rhythm.

The skill here shifts from production craft to editorial judgment. Knowing what to cut matters more than knowing how to build a slide.

The manager enabler

The third role works with the 1,600 or so frontline managers who supervise those 40,000 workers. Their job is to make sure managers use the data: that the morning huddle covers the three people who failed yesterday's check, that the new supervisor knows how to coach rather than re-lecture. We argued in the manager as single point of failure that direct-to-worker channels free managers to coach. Somebody has to teach them how. This role also owns the hands-on skill validation programme, because that's where humans remain essential.

Our guides on training frontline managers to coach and AI coaching for frontline managers are, in effect, this person's job description.

What happens to everyone else

I won't pretend the maths is comfortable. If a team of ten becomes a team of three, seven people are doing something else. In the companies I've watched go through this, the pattern has been:

• Content producers who were good at the craft move into the content-owner role or into the business, where their knowledge of the SOPs is more valuable than their slide skills.
• Logistics coordinators move into operations or HR, often into the workforce-management functions that still need scheduling.
• The people who were chasing completion become the analysts, because they know the data better than anyone.

Some leave. That's real. I'd rather say it than dress it up. What I'd push back on is the idea that the alternative, keeping a ten-person team doing production and chasing, is kinder. It isn't. It keeps skilled people doing work that a machine does better, and it keeps the frontline waiting weeks for content that should take an afternoon.

The trap: automating the old model

The failure mode I see most often is companies buying AI authoring and then using it to produce the same 45-minute modules, faster. Or buying WhatsApp delivery and then sending the same quarterly compliance blast, to more people. The tool changes; the operating model doesn't; the team stays at ten because the work has been reshaped rather than removed.

The tell is when the L&D calendar still looks like a course catalogue. The 2028 team doesn't run courses. It runs a continuous stream: a lesson when the SOP changes, a check when the data says a site is slipping, a nudge when a worker hasn't completed something in a week. Our 10 signs your frontline training is broken is essentially a diagnostic for whether you've changed the tool or the model.

What this means for L&D leaders now

Three things I'd do this year if I ran a frontline L&D function.

Measure where your team's hours go. For one month, log time by the four buckets above. Most leaders are surprised by how little goes to design and analysis.

Pilot the collapse on one workflow. Take one SOP category, say food safety or returns handling, and run it end to end the new way: AI-authored, phone-delivered, dashboard-tracked. Measure time-to-publish, completion, and score. Compare with the old way. Our customer results page shows what those comparisons have looked like for others.

Start growing the analyst. Whoever on your team is closest to the data is your 2028 team lead. Give them the dashboards now.

Where we could be wrong

AI output quality. Today's AI authoring produces a good first draft from a clean SOP. It produces a mediocre one from a messy, table-heavy, jargon-filled document, and it can be confidently wrong. The review step is not optional, and in regulated content it may be substantial. If quality plateaus, the production bucket shrinks less than I've suggested.

Hands-on skills. For manufacturing, healthcare, construction and logistics, a large share of critical skill is physical. Those functions will keep more trainers on the floor, and rightly.

Organisational appetite. Many companies will keep the ten-person team and the course catalogue for years because change is hard and nobody gets fired for running compliance the old way. My timeline may be optimistic.

I'm selling the tool that does this. Read accordingly. But also ask your current team the time-log question, and see what they say.

FAQ

Will AI replace L&D teams?

No. It replaces the production and administration work that consumed most of their time. The design, analysis and manager-enablement work that remains is the part L&D was always meant to do.

How fast can AI turn an SOP into a lesson?

On Leap10x, a PDF, PPT or Word document becomes a sequence of micro-lessons with quizzes in under fifteen minutes, translatable into 70+ languages with one click. A human should review before publishing.

What ratio of L&D staff to frontline learners is realistic?

We've seen small teams run tens of thousands of learners once content, delivery and tracking are automated. The exact number depends on how much of your training is hands-on.

What skills should frontline L&D people build now?

Data analysis, editorial judgment, and manager coaching. Less production craft, more deciding what matters.

Does Leap10x replace an LMS?

For frontline knowledge training, communication and AI assistance, it replaces the parts of an LMS that frontline workers never used. For long-form certification courses or desk-based staff with email, a traditional LMS may still fit; we say so on our own site.

---

Want to see how long your worst SOP takes to become a lesson? Send it to us and we'll show you the output, in two languages, the same day. Book a demo or write to hello@leap10x.in.