It's 8:40 on a Tuesday and a hospital rep is sitting in a car park, twenty minutes before a meeting with a cardiologist who, last time, asked a question she couldn't answer. She has a tablet full of approved materials, an LMS login she last used for a compliance module in September, and a cycle meeting deck she half remembers. What she wants is not a course. She wants to ask one question and get one trustworthy answer, then walk in.
That gap, between where knowledge is stored and where it is needed, is why workplace learning is shifting from training events to conversation. Learning is moving from a place you visit to a thing you ask. For pharma and MedTech field teams this is a real opportunity, but it only works if the conversation stays inside the label, uses approved content and still builds memory rather than replacing it. This piece covers what that shift looks like, where it helps, and where it can go badly wrong.
The portal model assumed people have time to go somewhere
For two decades the default model of corporate learning has been the same. Stop what you are doing. Log in. Find the module. Complete it. Pass the quiz. Go back to work.
That model made sense when learning was mostly formal and scheduled: induction, product training before a launch, the annual compliance refresh. It still makes sense for some of that. But it rests on an assumption that doesn't hold for most customer-facing people in Life Sciences, which is that they have spare, uninterrupted time and the inclination to spend it in a learning platform.
They don't. Josh Bersin, writing in 2018 about what he called learning in the flow of work, cited older research suggesting the average employee had about 24 minutes a week for formal learning. It is a dated figure about formal learning time, not a measure of all learning or of every rep today, but it points the right way: for a rep juggling territory plans, call targets, CRM call notes, ride-alongs and the next cycle meeting, protected learning time is scarce.
So the modules get done in a rush, usually the week before a deadline, and the content fades. We've written before about the forgetting curve. Murre and Dros's 2015 replication of Ebbinghaus, a single-participant relearning experiment with nonsense syllables, shows that unrehearsed material fades fast (savings of about 4% after 31 days) and that spacing helps. It can't tell you how much of a launch training survives to week six, but it is a good reason not to rely on week one alone.
The problem was never access to content. Every rep we've worked with has more approved material than they could read in a year. The problem is distance: the distance between the knowledge and the moment it is needed.
No destination, not no screen
You may have heard this trend called "screenless learning". The label is confusing, since most of the examples below still run on a phone. Nobody is taking away the screen. The point is to remove the behaviour of entering a separate system to learn.
The old sequence goes: need information, open a platform, search, consume, return to work. The new sequence is shorter: need help, ask, get guidance, act.
In practice that help can take several forms:
- a knowledge assistant that answers questions from approved documents
- short practice challenges that take a minute and happen between calls
- audio that can be listened to in the car or on a train
- a manager's coaching nudge after a ride-along, pointing to one specific topic
- a quick self-test the evening before an important meeting
The format matters less than the behaviour. Learning stops being a destination and becomes part of the working day, which is exactly where field teams spend their time.
Why conversation is the shift that matters most
People now expect software to answer them. Two years of widely available generative AI have changed expectations faster than most learning teams have changed their platforms. Nobody wants to browse a catalogue of content tiles when they could type the question.
Think about the questions a rep actually has in the week before a key account meeting:
- "What is the approved way to answer the question about renal dosing?"
- "How do we position against the competitor's new data, within our label?"
- "Test me on the subgroup results before tomorrow."
None of those is a request for a course. They are requests for performance support, and some of them are requests for practice. That is a different job from content delivery, and it asks different things of the technology behind it.
It also changes the role of L&D and Commercial Excellence. When learning is conversational, the team's job shifts from publishing modules to curating a trustworthy body of approved answers, deciding what gets practised, and watching what people ask.
The problem was never access to content. The problem is the distance between the knowledge and the moment it is needed.
In pharma, a conversation has to stay inside the label
Here is where the generic enthusiasm for conversational learning runs into the reality of a regulated industry.
A general-purpose chatbot that draws on the open internet, used without the controls below, can answer a question about off-label use, cite a study your medical team has never reviewed, or produce a percentage that sounds right and isn't. General tools are improving quickly, so treat this as a question of set-up and governance, not a fixed limit of the technology. In a retail setting that is embarrassing. In front of an HCP it is a compliance incident.
So a conversational layer for a field force needs a few non-negotiables:
- It answers only from content that has been through medical-legal review, not from the wider web.
- It says clearly when the answer is not in the approved material, rather than improvising.
- It routes medical questions outside the promotional scope to the right place: an MSL, medical information, or the local medical team.
- It keeps a record of what was asked, so Medical Affairs and Marketing can see where the field is uncertain.
That last point is underrated. The questions reps ask a knowledge assistant are a direct read on where the messaging is unclear, where the visual aid is missing something, or where a competitor's claim is landing. In our experience, what people ask tells you almost as much as what they get wrong.
Asking is not the same as knowing
A fair objection to all of this is that if a rep can ask for any answer at any time, why bother making them remember anything?
Because the moment that matters is not in the car park. It's in the room, three minutes into a conversation with a specialist who has just raised the exacerbation data or the safety profile in older patients. The rep needs the core answer at hand, and can check approved materials or an approved support tool when a detail needs confirming. Confidence in that moment comes from retrieval practice, from having recalled the right answer several times before, under a bit of pressure.
That is why conversational support and structured reinforcement are complements, not competitors. The assistant handles the long tail of occasional questions. Short, spaced practice builds the core knowledge that has to be available instantly. Gamified challenges, played in a minute between calls against colleagues, do the second job well because people come back to them without being chased.
In the programs we run, the pattern is consistent. Content on its own moves knowledge a little. Repetition in short bursts moves it more. And a small set of stubborn misconceptions only moves when a manager or medical colleague sits down and explains why. A conversational tool doesn't change that hierarchy. It adds a useful layer at the top.
Audio is the most overlooked format
If there is one format we think learning teams are underusing, it's audio.
Field teams spend a surprising amount of their week driving between accounts. That time is almost never used for learning, because reading while driving is obviously out and most e-learning is designed to be looked at. Yet people already listen to podcasts, voice notes and audiobooks in exactly those moments.
Short audio pieces fit field life well: a ten-minute episode on the new guideline before a round of respiratory calls, a product update ahead of the cycle meeting, two experienced colleagues talking through how they handle a common objection. With the new generation of AI voice tools it is becoming realistic to produce this kind of content from approved documents quickly, provided the scripts go through the same review as any other material.
Audio is not a replacement for practice, though. Listening provides content; follow-up questions can check what the listener can recall. It works best as a primer that is followed by a short quiz or challenge on the same topic, so the listener actually retrieves what they heard.
Structured learning isn't going anywhere
None of this means the end of formal training. Launch training, certification on a new device, pharmacovigilance and compliance modules, onboarding for new hires: these need structure, sign-off and a record. Regulators and auditors will keep asking for evidence that people were trained.
What changes is the balance. Structured learning provides the depth and the record. Conversational support and short reinforcement keep that knowledge alive and usable for the months between events. A launch plan that puts all of its effort into the kickoff and nothing into weeks two to twelve is betting against the forgetting curve, and the forgetting curve usually wins.
The practical question for Commercial Excellence and L&D is therefore not "should we get rid of the portal?" It is: which business problems does conversational support solve better than a module? For compliance administration, probably few. For launch readiness, objection handling, product updates between cycle meetings and onboarding new reps into a territory, quite a lot.
Where this leaves field learning in 2025
Our view, heading into the new year, is that the winning set-up for a field force will look like this: a small number of structured events tied to real business moments, such as the kickoff, a launch or a congress; short, gamified reinforcement in between; a knowledge assistant that answers from approved content; and increasingly, audio for the time people spend on the road.
The common thread is less friction. Reps don't want another portal, another login or another dashboard. They want to ask, get a trustworthy answer and get back to their customers, and they want to feel sure of the key facts before they walk in.
At Atrivity we have spent years on the practice side of that picture: one-minute challenges, battles around business moments and follow-up that sends each person the fix for their own mistakes. We pair that with Atrivity Assistant, based on approved documents. If you're planning your 2025 launches and wondering how learning fits into the flow of field work, a good question for your next planning meeting is this: when a rep has a question at 8:40 in a car park, where do they go today?