Section 1: Learning used to be part of the job
When I started my working life, learning was something an organisation did with you, not something you were left to sort out on your own. There was structured, classroom-based training, and there was a budget for it. If you could identify something you needed to learn to do your job better, you could usually make the case, and more often than not the money was found. Most importantly, there was time. Learning sat inside the working week rather than being squeezed around its edges. Nobody saw this as a perk. It was simply how organisations made sure their people knew what they were doing.
My name is Connie Hazendonk. I’m a Salesforce transformation consultant and enablement specialist. For more than ten years I’ve worked with organisations across sales, service, field service and higher education, from mid-market businesses to global enterprise rollouts, and I’ve trained thousands of users in more than a dozen countries. Almost all of that work comes back to one question: why don’t people use the systems their organisations have invested so much in? The answer is rarely the technology. Far more often, it’s what happens, or doesn’t happen, around the people expected to use it.
That doesn’t mean everything used to be handed to us. There was no online learning of the kind we take for granted today: no free platforms, no video tutorial for every feature, no community forum with the answer waiting for you. If you wanted to go further than the classroom took you, you did your own research. You read the manual, found a book or asked the person at the next desk until you understood. But that self-directed learning sat on top of a foundation the organisation had deliberately put in place. You knew enough to know what you didn’t know, and that’s what made the rest of the learning possible.
Somewhere along the way, that foundation disappeared, and I don’t think anyone actually decided it should. The training budget was trimmed. The classroom became a webinar, the webinar became a link, and the link became a line in an all-staff email saying resources are available for anyone who wants them. Each step felt reasonable and each step saved money. By the end, responsibility for learning had quietly moved from the organisation to the individual, and the time that used to come with it had gone too.
The irony is that there has never been more learning available. Every major technology company now offers free courses, many of them genuinely excellent. A motivated person with a free evening could teach themselves an enormous amount. What has gone isn’t the content. It’s the structure, the time and the sense that learning is part of the job rather than something you do when you get a chance. For most people, that chance never comes.
Now AI arrives, and of all the technologies we’ve introduced into the workplace, it is the one where this matters most. Leaders tell their people to be curious, to experiment, to explore and to take responsibility for their own development. It sounds empowering, and it treats people as capable adults rather than passengers. But it’s the second half of my story without the first. It’s the self-directed learning without the foundation it used to stand on.
What strikes me is that almost everyone who uses AI well learned it in exactly the way we’ve stopped offering. Anyone can sit at home with a laptop, open a tool and start playing with it. Very few people learn much that way on their own. They learn by pushing and testing it alongside other people, comparing what they got, arguing about whether it’s right and trying again. Learning happens together. You have to be in the room.
I’ve seen this in every classroom I’ve run. Adults learn to use technology in their work by watching a colleague do something they didn’t know was possible, by asking the question they’d be embarrassed to type into a search box, and by trying something with real work in front of them, getting it wrong and having someone nearby to explain why. The technology companies know this too. Several have opened physical spaces where customers and partners come to learn AI together, in person, because they understand that’s where the real learning happens. That raises an uncomfortable question for the rest of us. If the people building these platforms invest in rooms, groups and guides to help others get the most out of AI, why are we rolling it out to our own people with an email and a link?
Part of the answer is that self-service looks very good on paper. It produces numbers. You can report how many people have access to the learning platform, how many modules were completed and how many badges were earned. What those numbers rarely tell you is whether anyone can now do their job differently. A person who has been sent a link counts as someone who has been offered training. A person who watched a webinar with the camera off while answering emails counts as trained. Calling that training is generous. It’s exposure, and exposure has a way of feeling like progress right up until someone has to use the tool with a customer waiting.
Then there’s the question of who actually has the time to make the active choice. Think about the people at the edges of most organisations: the field engineer driving between jobs, the service agent measured on how quickly each case is closed, the sales rep whose target resets every quarter. These people aren’t short of curiosity. They’re short of slack. Every minute spent experimenting with a new tool shows up somewhere in their numbers. Many of them are also quietly unsure what they’re allowed to do. Can I put customer information into this? What happens if it gets something wrong and I send it anyway? Whose fault is that? When the rules are unclear and the stakes feel personal, the rational choice isn’t to experiment. It’s to wait.
People who wait rarely announce it. They don’t complain about the rollout or fill in a feedback form saying they feel left behind. They carry on working the way they always have, and their silence is easily mistaken for a lack of interest. It isn’t. Silence is usually uncertainty that nobody has asked about.
None of this means personal responsibility doesn’t matter. It does, and nobody becomes good at anything purely by being taught. But personal responsibility works best as the second layer, not the first. It’s what people build on top of a foundation the organisation has already put in place. Remove the foundation, call what’s left empowerment, and you haven’t created a culture of learning. You’ve simply created a way of finding out who would have learned anyway.
Section 2: The divide we keep recreating
Watch almost any organisation start using AI and the same picture appears within a few months. There’s a small group using it constantly, several times a day, and getting real value from it. There’s a middle group dipping in once or twice a week, mostly for the obvious things. And there’s a large group who have barely touched it at all. Commentators have started calling this a new digital divide, and leaders are rightly worried that it will widen.
What I’d question is the word “new”. We could have had this exact conversation five years ago, or ten. We’d have described the same categories: the digitally confident, the frozen middle and the people who simply aren’t engaged. The technology is different. The divide looks the same.
I’ve watched that shape appear inside Salesforce organisations for more than a decade, and it has nothing to do with AI. Every rollout I’ve been part of produces it. There’s a confident group who build their own reports, find the shortcuts and quietly become the people everyone else asks for help. There’s a middle group who do what they’ve been shown and not much more. And there’s a group who do the minimum the system demands, while keeping the real information somewhere else entirely: in their heads, in their notebooks or in the conversations they have in the car park.
What rarely gets talked about is what happens to that first group. Because they’re confident, they get volunteered. They test the new release before anyone else sees it. They sit in user acceptance testing. They train their colleagues, act as champions during the rollout and become the unofficial first line of support long after go-live. Very rarely does anyone take anything off their plate in return. These are usually also the strongest performers in their teams, so their day job is already the most demanding. Over time, the people the whole rollout depends on become its bottleneck. Some burn out. Some quietly stop volunteering. Some leave and take everything they knew with them. The organisation mistook their goodwill for a strategy.
Now the same people are being volunteered again, for the AI pilot, the prompt library and the lunch-and-learn. The irony is hard to miss. The people best placed to help everyone else are given the least time to do it properly.
Leaders rarely see the rest of the picture clearly because they tend to see it through that confident group. They’re the ones invited to the demo and the ones whose results appear in the quarterly review. The average looks healthy because the best performers are pulling it up. The real story is the gap between your most confident team and your least confident one, and very few organisations measure it.
This is the part that I don’t think is being discussed honestly enough. AI doesn’t arrive on a blank page. It arrives on top of the habits, the confidence and the information that already exist. The people who were confident with the CRM tend to be the people who will be confident with AI, because many of the skills involved are closely related.
I’ve argued in earlier episodes that AI changes the skill people need, from knowing where to click to knowing what to ask and how to judge the answer. I still believe that, and I don’t think it contradicts what I’m saying here. Some skills carry over and some become far more important. What carries over is understanding: what the system actually holds, where the data is weak, how the process really works rather than how the diagram says it works, and what this particular customer is really like. What fades is navigation, the clicking and the knowing which tab to open. And judgement, which was always part of the job, moves from the edges to the centre. Judgement doesn’t come from nowhere, though. It’s built on exactly that understanding. So the people who were confident with the CRM have a head start on the skill that now matters most, while everyone else is being asked to judge answers about a system they never really understood.
Picture what that judgement looks like in practice. An experienced sales director opens her AI assistant on a Monday morning. It has already pulled together everything that needs her attention and prepared a briefing for every meeting in her diary. The lazy option would be to rely on it completely. Instead, she writes down the three priorities she wants her team to focus on this week, then asks the AI what she’s missing, where she might be wrong and what she has overlooked. She uses it to challenge her thinking rather than replace it.
That’s judgement, and it’s built on years of experience and a deep understanding of her business. When the answer doesn’t feel right, she has a basis for noticing. Now picture a sales rep three months into their first role, opening an AI summary of an account they inherited last week. It’s fluent, well structured and completely confident. On what basis are they supposed to argue with it?
This is what changes when AI enters everyday work. People stop being the authors of their records and become their editors. Editing sounds easier than writing, but it isn’t. To edit well, you need to understand the subject better than the draft does. You need to know what’s missing, what’s out of date and what simply doesn’t sound like this customer. That takes more understanding, not less.
There’s a second reason the divide compounds, and it sits underneath the technology. Every serious case for AI in business rests on grounding it in a company’s own data, and that data had better be organised and it had better be good. Every Salesforce customer nods at this. Very few stop to ask where good data comes from. It’s the residue of thousands of small decisions made by busy people: the deal left open because marking it lost felt like admitting failure, the call that was never logged because logging it took longer than the call, the duplicate record created because someone didn’t trust the search. Every one of those decisions is a behaviour, and every one of them is now material for an AI that will read it, summarise it and present it back with complete confidence.
So the divide doesn’t reset when a new technology arrives. It compounds. The confident group accelerates and carries more of the load, the middle stays where it is, and the gap grows wider, faster and with far less visibility than before. AI adoption is downstream of CRM adoption, not a shortcut around it. If you never closed the gap with the system you already own, you aren’t starting your AI journey from zero. You’re starting it from wherever your least confident team happens to be.
Section 3: Who’s standing behind the door
If self-service learning mostly rewards the people who were already ahead, and if AI deepens a divide most organisations never closed, a reasonable leader will ask what they’re supposed to do on Monday morning. They can’t wait for the education system to catch up, and most can’t build an AI centre of their own.
When people talk about closing the AI skills gap, the conversation usually turns to access: more free courses, better signposting, a single front door where anyone can find the right learning at the right time. That’s useful, and it addresses a real problem. Plenty of free learning exists, and most people have no idea where to start. But inside an organisation, the front door question is both simpler and harder. It isn’t where the resources are. It’s who is standing behind the door.
Before we get to who, we need to be honest about priorities. Every person in your organisation has targets, a full diary and only so many hours in the day. They’ll spend those hours on whatever the organisation measures, rewards and asks about. If learning isn’t prioritised by the business, with time protected, expectations set and managers following up, it won’t happen, however many resources are available. That isn’t a lack of willingness. It’s a perfectly sensible reading of what the organisation actually values. Every leadership team says learning matters. The diary tells you whether they mean it.
We also need to be honest about how AI is actually learned, because it isn’t learned by watching a few videos or being handed a list of prompts. A list of prompts is a bit like a phrasebook. It will get you through a simple exchange, but the moment the conversation goes somewhere unexpected, you’re lost. In my experience, people learn fastest and best in classroom-style sessions where someone guides them through realistic scenarios from their own work, and where they can pull the output apart, experiment with it and discuss it with the people around them. Why did it say that? What has it missed? Would you send this to your customer? That discussion is where judgement gets built, and it’s very hard to build on your own.
So who stands behind the door? Start by making someone own how AI is used once the licences are bought and the launch announcement has gone out. In my experience, the rollouts that work best are rarely the ones with the biggest technical teams. They’re the ones with a determined owner from the business, someone who cares whether it works for the people doing the job and for their customers, not just whether it has been switched on. Most organisations can name the person who owned the build. Almost none can name the person who owns the six months after. A simple test is to ask who would notice if people quietly stopped using AI next quarter, and whose job it would be to do something about it.
The second move is to train around the moments of work rather than the features of the tool. A rep doesn’t start their day wanting to explore AI. They start it with a customer call in an hour and a pipeline review on Thursday. A service agent has a difficult case and a queue behind it. The useful question is never what the tool can do. It’s what good looks like in this moment, now that AI is part of it. Show people how to prepare for that call, what a good account summary contains and what it usually misses, and do it together, with real work in front of them.
The third move is to teach the question that experienced sales director asks herself: what am I missing, and where am I wrong? It’s the most valuable AI habit anyone can have, and it can be taught. Give people outputs that are subtly wrong and ask them to find the problem. Make questioning the answer a normal, expected part of the job rather than a sign that someone doesn’t trust the technology. And recognise that whoever writes the instructions behind your AI tools, the templates and prompts that tell them what to summarise and what to prioritise, is now shaping how your people think about their work. Whoever writes your AI skills is writing your training.
The fourth move belongs to managers, and it may matter more than all the others. Users don’t follow PowerPoints. They follow their managers. If a team leader opens the Monday meeting by asking the AI summary what it got wrong, the team learns that questioning is expected. If the same leader forwards AI-generated updates without reading them, the team learns that too, and no amount of training will undo it. Enablement isn’t just what you teach. It’s what you tolerate.
The fifth move is to protect the confident few. If you’re going to rely on them to test, train and support everyone else, make that part of their job rather than an addition to it. Give them the time, take something off their plate and recognise the work properly. Otherwise you’ll burn out the very people your AI rollout depends on.
The final move is to watch the quiet ones. Stop reporting AI adoption in terms of access, logins and completed modules, and start asking people how confident they feel using it in their actual work. Pay attention to the team that never asks questions, because a team with no questions is rarely a team with no problems.
None of this needs a new building, a national programme or a large technology budget. It needs a decision that the organisation, not the individual, is responsible for making sure people can use the tools it has chosen.
When I started my career, nobody thought giving people time to learn was a luxury. It was simply how you made sure they could do the job. AI hasn’t made that less true. It has made it urgent, because the technology now arrives faster than ever, just as the time to learn it has quietly disappeared. So the question isn’t whether your people are curious enough to learn AI. It’s whether your organisation still believes learning is part of the job.
If your organisation wants people to use AI and Salesforce well but hasn’t given them the time or structure to learn, that’s exactly what I help fix. You can find out more about how I work with teams to build training people actually use here: sebapp.com
Be the first to comment