One of your learners is halfway through an activity and they aren’t really sure what exactly it’s asking for. It’s not a crisis, and they’re not trying to ‘game the system’. They’re just… a bit stuck.
Maybe the wording of the activity is unclear. Maybe they are unsure what kind of evidence would be acceptable. Maybe they have read the task three times and still cannot work out where to begin. Their tutor is in back-to-back reviews, or off for the evening, or dealing with something more urgent. So the learner waits, or guesses, or sends a message that their tutor doesn’t see until tomorrow or the next day.
This is how progress slows down in real life, not because the learner is incapable or the tutor has failed. Because apprenticeship and other funded training delivery depends – far too often – on the right person being available at the exact moment that a learner needs them, and most of the time, they cannot be.
Trainers are not short of care; they are short of available hours.
Your tutors and coaches are already carrying a huge amount: teaching, reviewing evidence, preparing learners for milestones, liaising with employers, checking progress, recording what needs to be recorded, and dealing with all the human things that sit around a long learning journey. Dips in confidence. Employer issues. Missed deadlines. Safeguarding concerns. Motivation that comes and goes.
And on top of those things come the smaller questions:
These are not silly questions; a lot of them are exactly the questions a learner should be asking. But when every one of them has to go back through a tutor, the process gets clogged up with work that is important, but not always the best use of human judgement.
And so the learner waits longer than they should, the tutor repeats themselves more than they should, and the questions that really do need a conversation get squeezed by the questions that could have been answered earlier, in a more structured way.
This is where AI could be useful. But only if we are honest about what kind of usefulness actually matters in funded skills delivery.
A generic AI tool can help with plenty of things: it can explain a concept, tidy up wording, suggest a structure, or give someone a way into a topic. Most people who have used ChatGPT or Claude for more than five minutes know that.
But when a learner asks, “What should I do next?”, the really useful answer depends on the programme they are on, which activity they are working through, what criteria are attached to it, and what progress they have already made. If they ask, “What does this mean?”, then an answer needs to relate to the actual task in front of them, not a broad explanation of the topic.
A blank chat box does not know any of that – the learner has to supply context: explain what they are studying, where they are in the programme, what the activity description says, what work they have already done, and what kind of help they need. Then they have to judge whether the answer they get is actually relevant and suitable.
That’s a lot to ask of someone who is already unsure!
Generic AI can help some learners, some of the time.
It may also take them outside the platform, outside the flow of their work, and into a space where the answer sounds confident but has no real understanding of the actual work they are trying to complete.
And that is not a sustainable support strategy for funded learning, because the risk is not just that the answer might be wrong – it is that the answer might be nearly right. Plausible enough to follow, but not quite aligned to the activity. Helpful-looking, but not actually in line with the provider’s expectations. Encouraging, but nudging the learner in the wrong direction.
So AI support can’t be so vague or disconnected that it leaves learners guessing, but it also can’t be so helpful that it starts doing the work for them.
Evidence is meant to demonstrate what a learner understands, how they’ve applied their knowledge, and where there is still some work to do. This is so important in funded learning. AI that hands a learner something they can copy and paste doesn’t help the learner learn, it doesn’t help the trainer judge, and it doesn’t help the employer looking for real on-the-job competence.
The more useful, valuable role for AI is actually narrower.
Let’s use it to help the learner understand what an activity is asking for. Help them break down a task into bitesize chunks. Help them think about what evidence might be relevant. Let’s help them get moving again, without pretending that AI can replace the judgement of a tutor, assessor or coach.
It’s such an important distinction because tutors do so much more than ‘answer questions’. They notice patterns, understand each individual learner’s confidence, motivation and circumstances. They can tell when a simple blocker is not so simple anymore. They understand when to challenge, when to encourage, when to escalate, and when something needs proper human attention.
We shouldn’t be relying on AI to make those calls quietly in the background.
“AI assistant” is easy to say, but delivering useful learner support, built into the actual delivery environment, is harder.
That’s the challenge Bud Assist for Learners addresses.
Not another empty prompt box or chatbot, but a way for learners to ask for help from right inside Bud, where their activity, learning plan and recorded progress are already part of the picture.
So next time your learner gets stuck, they don’t have to start by explaining everything from scratch. They can ask for help, right there, in the same place that they’re trying to do the work. They can benefit from instant, always-on support that is shaped around the specific activity they are looking at, and receive the kind of guidance that helps them to get unstuck, without actually doing the work for them.
For providers, that means a better, smarter way to support their learners ‘through the gaps’.
We’re not building Bud Assist for Learners to make apprenticeship delivery less human, but to stop very ordinary and very solvable moments of confusion becoming bigger problems, just because no one was available in the moment.
Bud Assist for Learners is coming soon - don’t miss the first-look webinar on 15th July.