An adaptive learning layer for NGL’s collaboration with the Harvard Data Science Initiative
Güneş Ozgün, Product Engineer
There is an exciting new model emerging for AI education: one that combines the energy and human connection of learning together with AI-augmented learning that can adapt to each person.
Next Gen Learning (NGL) builds and runs intensive courses for leaders driving AI change in their organizations, in collaboration with the Harvard Data Science Initiative. The core learning path combines the best of human teaching with the best of personalized learning. Students learn through synchronous lectures and labs with live faculty, then get hands-on support from AI implementation experts in office hours. From there, they build their own AI playbooks with NGL’s AI tutor and deepen their technical knowledge through Scholé’s adaptive curriculum.
Between live sessions, Scholé gives participants space to revisit what each intensive teaches, deepen their understanding, and apply it in the context of their own work. Topics range from RAG to MCP, multi-agent systems to LLM architecture intuition, Claude Code to responsible AI. Dr. Vinitra Swamy, Scholé’s CEO and co-founder, is among the teaching faculty for the intensives.
- 2,000+
- leaders and executives learning with Scholé across four NGL intensives
- 11,000+
- lessons opened by participants between live sessions
- 9 / 10
- average likelihood of recommending Scholé
Participants join from every continent, which is precisely why a fixed syllabus cannot serve all of them equally.
Extending the intensive beyond the live sessions
NGL builds the core community, and Scholé helps support the outliers: bringing participants to a shared starting point with pre-work, then augmenting NGL’s live sessions with technical deep dives and constantly updated, tool-specific knowledge. For each course, we co-designed a curated lesson path that builds on what the cohort is learning, so participants can revisit technical ideas, practice them through real-world tasks, and push further in the context of their own work.
Today, four intensives run on Scholé:
- Agentic AI
- AI Strategist and Builder
- AI Leadership: Strategy, Governance, and Agentic Systems
- OpenClaw Strategy
“Working with Vinitra and the Scholé Team has been hugely enabling for us. While we focus on the core program learning path that combines live faculty sessions with our AI tutor, Paski, students break out into adaptive sessions with Scholé’s AI tutor to close gaps, deepen understanding and so much more. It’s a pleasure to work with the Scholé team.”
What they used
- Curated course paths
- Open catalog
- Written answers with AI feedback
- Podcast mode
- Olé tutor
Scholé gets more personal as someone learns. Onboarding asks one thing: what industry the participant works in. Everything after that is earned rather than declared. Each finished lesson adds to their profile, and Scholé adapts from what they actually do, how they answer, what they find easy, how they like to learn, and where they go next.
Throughout, participants work with a cast of pedagogical agents led by Olé, Scholé’s metacognitive tutor. Olé introduces ideas, asks questions, and adapts as they learn, providing AI support that is hyper-personalized to each learner rather than identical for the whole room.
Two features earned disproportionate use from this audience.
Written answers with AI feedback. Instead of multiple choice, Scholé puts an applied scenario in front of someone and asks what they would do. Olé reads the response, weighs it against the concept and explains what it would add. Participants wrote 433 of these in a month, answering with their own approval thresholds, reporting systems and compliance requirements. Olé’s feedback responds to the specifics of each answer, not to a model answer.
Podcast mode. Scholé generates an audio version of a participant’s own material, so a lesson can be listened to rather than read. Over 50 podcasts were generated as soon as the podcast feature was soft-launched during the course, by 31 early adopters fitting study around a commute and a full calendar.
“It is too valuable a site not to have access to it, and go through the materials at my pace.”
What they chose
The curated path is a recommendation, not a fence. Behind it sits Scholé’s public catalog, 105 lessons across ten categories, and every participant can reach all of it.
They used that access heavily. Between October 2025 and August 2026, beyond the full learning path their courses assigned, 136 participants went looking for something nobody had asked them to study, and between them opened 86 additional lessons from the catalog.
What executives chose for themselves
Lessons participants opened from the wider catalog beyond the full learning path their course assigned, grouped by category.
Lessons opened between October 2025 and August 2026 that no course path assigned: 86 lessons, 136 participants, 479 openings in total. Excludes the starter lesson every learner lands on at signup, and lessons participants generated for themselves.
Two categories led: AI Tools and Data Science.
AI Tools drew the most: AI-enabled design in Claude Design and Figma Make, advanced spreadsheets, workflow automation in App Script, APIs, running models locally, verifying code that an AI wrote. Data Science came next: the lifecycle, visualization, querying data with SQL and LLMs, exploratory analysis, time series, preparing data for models.
This is the behavior of senior people building the layer underneath the strategy they were taught. An agentic AI framework is only as good as your ability to judge what it is running on, and these participants went and got that judgement themselves.
That matters because any program is scoped deliberately, and no syllabus can carry every direction a room of senior people might need to go. The catalog gives each participant somewhere to take the learning next.
This audience already has a starting point. Participants arrive from a live session holding a framework, then use Scholé to take it somewhere more specific. The curated path gives every cohort a shared foundation; the open library lets a participant in insurance and a participant in local government go in completely different directions from the same lecture.
“I’m now at Level 3 in Scholé and intend to go all the way to the top. I’m thoroughly enjoying it.”
What participants said
Participants rated lessons 4.4 out of 5, with 81 percent of ratings at four stars or higher. Asked how likely they were to recommend Scholé to someone else, they averaged 9 out of 10.
“Love what you all are doing at Scholé! FINALLY a learning system that adapts to the learner!”
“Congratulations on Scholé AI, such an elegant application of agentic capabilities for personalized, role-based learning. Truly impressive!”
Live teaching will always be a gold standard, and pairing it with an agentic tutoring system that meets each person where they are creates a world-class learning experience. If you look at the photos above, you will also notice how truly global this initiative is. Learners and faculty join from different geographies, time zones, and backgrounds, all with the same enthusiasm to be in the virtual room together.
That is the shape of what we are excited to keep building: a strong shared learning experience, a curated path that mirrors what you teach, an open catalog for exploration, and a tutoring system that lets each person learn in the way that works best for them.
What a pleasure it is to work together. So much gratitude to Sam, Hagen, and the NGL team; Xiao-Li and the HDSI teaching faculty; Dirk, Ulla, and the DAIN Studios team; Stephanie, Miguel, and Ani; and, most of all, the many learners who choose to spend their time learning with us.