Bubot Learning Platform

Ed Week ran a piece last year on Khan Academy's AI tutor. The most common student response, reportedly: "Bro, IDK."

Not a knock on Khan - they're doing good work. But it pointed at something real. When a learning environment asks a student to produce answers to content questions, the student often produces the minimum.

In our sessions, across 471 analyzed transcripts, the average learner writes 500 words per session. The longest conversation ran 83 turns and over 2,700 words. These are professionals practicing design thinking and risk management - voluntarily investing an hour of focused effort in a simulated scenario.

The difference isn't motivation. It's architecture. When the scenario requires you to interview a user, synthesize an insight, and defend a recommendation, "IDK" isn't an option - the scenario won't advance. You have to think.

That's the whole design brief. Build an environment where the only way through is to practice the methodology.

Curious what your team's transcript would look like.

"What have you done in this situation? That reveals actual competency. What would you do is just hypothetical."

A talent executive at a nonprofit said that to us. She's been hiring for twenty years. She's describing the single biggest gap in how we evaluate workforce readiness - we ask people what they'd do, and we call the answer evidence.

It isn't. A capable person and a confidently-incompetent person give the same answer to a hypothetical. The audit fails when the actual moment arrives.

This is why we built Bubot around behavioral evidence instead of assessment scores. Every session produces a transcript - how the learner reasoned through the scenario, what questions they asked, where their judgment was strong, where it broke down. That's evaluable. A quiz answer isn't.

"What have you done" is the question every hiring manager, regulator, and performance reviewer is actually trying to answer. Our data should help them answer it.

What does your training currently let you claim - and what would you rather be able to claim?

86% of traditional assessment questions test recall - the lowest level of Bloom's taxonomy. Remember the rule. Recognize the term. Pick the correct answer from four options.

Under 15% test higher-order thinking - analyze, evaluate, create. The cognitive work that actually matters when the situation isn't in the textbook.

In Bubot sessions, 48.6% of learner turns demonstrate higher-order thinking. Three times the traditional rate. Not because learners are smarter. Because the scenario forces them to apply judgment, not retrieve facts.

This is the quiet reason completion metrics feel hollow. We've built an entire measurement infrastructure around the easiest thing to measure, and the hardest thing to measure is the thing organizations actually need.

Bloom's pyramid isn't academic. It's an audit trail. 48.6% vs 15% is the difference between "your team passed the test" and "your team can reason through the problem."

Which one does your current training data actually show you?

500 words per session. That's what a Bubot learner writes, on average, while practicing design thinking or risk management.

For context: Khan Academy's AI tutor frequently gets "Bro, IDK" as a response. The industry benchmark for self-paced eLearning is around 10-15% engagement. Ours is 90.7%.

The difference isn't content quality. It isn't AI model choice. It's architecture.

When a scenario requires a decision with consequences - interviewing a skeptical user, defending a risk plan to a stakeholder who pushes back - learners engage. When it requires clicking "next," they don't.

We built Bubot around that one observation. The 500-word number isn't a vanity metric. It's evidence. It's a transcript of how someone actually thinks through a professional methodology, which is exactly the thing a quiz can't show you.

If you want to see what it looks like - bubotlearning.com. Happy to walk you through it.

Here's what happens in a Bubot design thinking session.

A learner introduces herself to a user persona. Asks an open-ended question about their experience. Gets a surprising answer. Follows up. Reframes. Forty minutes later she's written 500+ words, extracted three real insights, and built a recommendation she can defend.

Nobody told her to practice empathetic interviewing. She practiced it because the scenario wouldn't move forward without it.

This is what "hard skills with embedded soft skills" looks like in practice. We don't teach communication. We teach design thinking - and you can't do design thinking without the conversation. We teach risk management — and you can't manage risk without navigating stakeholders. The soft skills aren't the product. They're the delivery mechanism.

The conversations training usually skips are the ones where the methodology actually lives.

If your team learns frameworks but struggles to apply them - the missing ingredient is almost always practice, not more content.

"Training happens in the nuance. It doesn't happen in the classroom."

That's an EVP at a major bank. A person whose teams face regulators, clients, and edge cases every day.

The training module can cover the rule. It can't cover the moment the rule meets a real client with a real problem and no obvious right answer. That moment is where the judgment call lives - and judgment is the thing organizations actually hire and promote for.

You can't read your way into it. You can't quiz your way into it. You get it by running the conversation, in some form, before the real one happens. Badly, a few times, with feedback.

That's the difference between training and practice. Training is the deck. Practice is the difficult conversation.

Which one is your team getting more of?

"95% completion - but how confident are you that they can apply it?"

A VP of L&D at a large tech company said that to us. She runs training for thousands of people. She reports completion numbers upstream every quarter. Everyone exhales when the numbers are good.

And she doesn't believe them. Neither do the other eleven leaders we interviewed.

The LMS did its job. People clicked through. They finished. The green number on the dashboard is accurate. The problem is that "finished" and "can do the thing under pressure" are not the same measurement - and we've built an entire industry pretending they are.

Nobody we talked to used completion as a meaningful signal. Every one of them raised it as a problem. That tells us something: the people responsible for training outcomes don't trust the primary metric we use to measure them.

What would you need to see instead? We're collecting answers.

A bank's chief operating officer said this to us, matter-of-factly, in an interview about compliance training:

"We say all the time - we just want to get through them."

Not "learn from them." Not "apply them." Get through them. That's the institutional posture toward mandatory training at a major financial institution, said out loud.

And she wasn't complaining. She was describing reality. The training is well-produced. The LMS tracks everything correctly. People finish. The dashboard is green. But the bar has quietly slid from behavior change to survival, and everyone in the org has adapted to it.

We interviewed 12 senior leaders across 12 industries this year. Every single one of them described some version of this, unprompted.

If you recognize this in your own organization, you're not alone. The question worth asking is whether we keep measuring survival — or start measuring whether anyone can actually do the thing yet.

86% of training assessments test the lowest thing a brain can do: recall.

12% of employees actually apply their training on the job.

We looked at those two numbers next to each other for a long time. They explain a lot.

If almost every quiz at the end of a module asks "do you remember the rule," and almost nobody is using the rule two weeks later, then what exactly are we measuring? Completion, mostly. We measure finishing. And finishing and learning have quietly become different things.

The interesting part: every senior leader we've interviewed this year knew this. Every one of them. They report the completion number up the chain and privately don't believe it.

If you run L&D and this feels uncomfortably familiar — what would you actually need to see instead of a green dashboard?

We call it training. Then we're surprised when it doesn't transfer.

Here's the thing we keep bumping into when we talk to COOs and heads of L&D: nobody believes the completion numbers they report. The training got finished. The dashboard is green. And the same people ask us, quietly, whether anyone can actually do the thing yet.

Training and practice are different words for a reason.

Training is what happens when you sit through a deck on design thinking. Practice is what happens when you interview a skeptical user, extract an insight, and defend the recommendation to a stakeholder who pushes back.

You can't quiz your way into judgment. You get it by doing the work, badly, a few times, with feedback.

That's what Bubot is for - design thinking, risk management, leadership. Hard skills where ex*****on needs human judgment. The conversations training usually skips.

If you're running L&D and your completion numbers feel like theater, tell us what you wish you could measure instead. We're collecting these.