Earlier this month, Disco hosted a group of forward-thinking CEOs and learning leaders for an off-the-record breakfast in New York, alongside Marc Metakis, McKinsey partner and founder of the McKinsey Academy.

‍

The question on the table: what does learning look like when AI is everywhere?

‍

What made the conversation work was the range of people at the table. We had an early-career lawyer figuring out which skills will last, an education leader building learning models for young people, McKinsey’s firsthand experience across the Fortune 500, a boutique consulting firm using Disco to teach others how to build AI operating systems, and a community leader whose members are CROs and GTM executives.

‍

Across these perspectives, a few through lines emerged. People want personalized AI learning experiences, and they still want to learn together. AI will become our most constant source of feedback. And everyone who builds learning, services firms and platforms alike, is about to be held accountable for outcomes.

AI is becoming our most constant feedback partner

Feedback sits at the heart of every real learning loop. The problem has always been scarcity. A coach or manager can only see so much of how you work.

When AI is present in the flow of work, that changes. It can see how you collaborate with colleagues and how you work with AI itself, then give you assessment and feedback in the moment. We agreed this will become commonplace, both inside programs and inside organizations.

From driver’s ed to GPS

One of the many ideas Marc shared that got heads nodding was that learning is moving from driver’s ed to GPS. Instead of a prep course removed from the work, people will get the right guidance at the moment they need it. When AI sits alongside the work, that in-the-flow guidance becomes one of the most immediate ways to add value for learners.

But anyone who has followed GPS turn by turn knows the catch: you arrive without learning the route. So the real question is how we help people keep their own judgment while following in-the-moment prompts. To extend Marc’s analogy, how do they keep their sense of direction?

We think part of the answer is each other.

People want personalization and each other

The technology now allows hyper-personalized, adaptive learning. It’s tempting to optimize for that alone.

The room pushed back. Cohorts of people with different perspectives and ways of working spark learning that a perfectly tailored path can’t. What we heard, clearly, is that people want both: personalized AI and adaptive experiences, alongside cohort and community learning.

Value is moving to outcomes

We spent a good part of the morning on one question. Where does the value in learning actually live: inputs, process, outputs, or outcomes? In practice, that’s the difference between content, sessions, completions, and people actually working differently.

We landed on outcomes. That’s what we’re on the hook for. The edge comes from pairing AI with human judgment and taste, which AI can’t supply on its own.

This has big implications for services firms, which will be held more accountable for results. The same is true for platforms. Providing the container is table stakes. We have to help deliver the outcome.

The best programs blend shared and personal

A few patterns showed up across very different contexts:

• Projects are the unit of learning. Some are individual and personalized. Others are team projects. Both matter.
• One curriculum, personal starting points. The consulting firm teaching AI operating systems runs a shared curriculum because cohorts reveal how others approach the same work. Personalized assessment then shows each person where to go deeper.
• Peers are the moat. For communities of senior leaders, the core value is learning with and from each other. Any redesign of the learning experience has to protect that.

Agents should make us better at learning together

If peers are the moat, the job of AI is not to replace the cohort. It’s to do the work that keeps a community alive, so people can spend their time learning from each other. Our test: every agent action should end in a human connection or a human contribution.

Here’s what that can look like:

• Matchmaking. Put the right peers together on day one, by role, goals and context, instead of hoping they find each other by week six.
• Facilitation. Seed the next conversation, summarize long threads, surface the question nobody has answered, and nudge the quiet voices in.
• Routing in the flow of work. When someone gets stuck in the middle of their real work, connect them to the peer who already solved it, or bring the problem back to the cohort.
• Community memory. Turn great discussions into knowledge the whole group can use, with credit going to the person who said it.
• Evidence of growth. Help given, work shared and peer feedback are some of the best signals that someone is actually changing how they work.

What we’re taking back to our partners and clients

The future of learning is personalized and shared, AI-powered and human-led, and measured by outcomes. If you’re thinking about the same shift, here are four questions to ask yourself:

1. Where could AI give your learners more frequent and more honest feedback?
2. Where are you trading away diverse cohorts in the name of personalization?
3. Are your agents bringing people closer together, or routing around them?
4. Are you measuring completions, or the outcome?

Thank you to everyone who joined our table. If you’d like to join future conversations like this, we’d love to host you. Reach out at sales@disco.co.