When we opened our recent webinar with Training Industry, we asked the audience what they hoped AI would do for L&D. Most of the replies came down to doing more with less.
Learning teams have been asked to scale faster for years, so that hope makes sense. Yet the two teams on our panel described a year that went well beyond output. Nikita Singh, Senior Leader of Learning and Certifications at Cisco, and Mazie Baskin, VP of Anaplan Academy, have each built AI into how their teams design and deliver training. Their results came from tackling problems the business already had, bringing their people along and keeping humans responsible for quality. Below are five things we took away from their stories.
1. Start with a problem the business already has
A year ago, many teams were still debating whether to use AI at all, or worrying that everyone else was ahead. The more useful question now is how to make it work in your own organization. Since every team is solving something different, every team starts from a different place.
Nikita’s team builds Cisco’s technical and certification training. Their first step was helping the team write better prompts. Over time they went further, moving to a content-as-code model, where the team writes courses the way software developers write code, in developer tools with AI agents working alongside them. They then built AI “skills,” playbooks that package the team’s style, brand and editorial guidelines, for editorial review, design, metadata and labs.
Anaplan’s challenge looked quite different. As Anaplan grew from a platform into an applications company with many more products, Mazie’s team had to move training away from “where do I click” toward how to implement well. So they mapped the competencies implementers need at each level and used AI to build training around them. Each team began with a problem it understood well, and AI gave it a new way to solve that problem.
2. Treat adoption as a people challenge
Ask teams what slows them down and most will name change management and readiness. The technology itself comes up far less. Adoption also spreads gradually, moving team by team and person by person as comfort grows.
Nikita was candid about how emotional this period has been for many people on her team, who wonder where their roles are heading. Her team frames AI as something that extends their expertise, and they’re putting the time it frees into new skills and more creative work.
Mazie starts with something more personal. She asks people what they love doing and where they want their careers to go. Once people get hands-on, the fear tends to fade and experimentation follows. One team member who used to score practical certifications wrote an AI agent to take over the scoring, and he now leads automation for the team.
As a learning partner that helps teams scale production, we also hear the same question from our own instructional designers. We answer the same way, by looking at how AI can make their work easier and their output better.
3. Choose tools task by task, and govern them closely
Much of the early talk about AI centered on finding the one right model. Teams with more experience have found that different tools suit different work, and that costs swing widely depending on the task and the model behind it.
Both panelists pick their tools deliberately. Mazie’s team uses Claude for configuration-heavy work and Synthesia for video. At Cisco, Nikita’s team judges each model on the quality of its output and on how many tokens it uses, which is what drives the cost of each task. Token usage hardly came up a year ago; today it shapes how her team designs prompts and agents.
Governance carries just as much weight. At Cisco, a governance committee approves tools, and Nikita’s team stepped back from voice cloning and digital avatars because of security and consent concerns. Anaplan runs every tool through a security review and checks content as it comes out. Mazie was frank that reviewing AI output closely is demanding work that needs someone detail-oriented.
Learners are watching for this as well. Many have doubts about the accuracy of AI-generated content and want to know a person has checked it. In regulated sectors such as pharma and financial services, scrutiny begins earlier, with questions about which models to use, where data goes and what good governance should look like.
4. Look beyond ROI
The learning teams at Cisco and Anaplan both produce training faster now. AI drafts first versions for SMEs and instructional designers to review and refine, and quality checks have become more consistent. For Mazie’s team, AI made it possible to get field readiness kits, enablement and certifications to partners in time for a wave of new product launches, which gave the whole company more agility.
Nikita added a useful caution. Steps such as publication don’t use AI yet, so gains in one stage don’t always carry through, and end-to-end ROI is still hard to see.
Some of the benefits teams care about most resist measurement. Greater agility, more room to experiment and the chance to put human expertise where it counts all came up in the discussion.
5. Reinvest the time in the learning experience
What excited us most was how both teams are using the time they’ve gained. Nikita’s SMEs had long wanted to build graded experiential labs, but grading every possible scenario was too complex until AI could help. Her team is also designing onboarding simulations in which AI agents play the boss and the customer while a new hire troubleshoots a network. Mazie’s team uses AI-generated video hosts as tour guides through the product, so technical training feels more like a guided experience.
Questions from the audience followed the same thread. People wanted to know how to make dry regulatory content engaging, how to spot skills gaps before training starts and how to show real change in performance. Each of those comes back to learner experience and business impact. That is where AI gives L&D the most room to grow, helping learning teams work closer to the business and solve its problems faster.
Watch the full conversation
The webinar recording goes deeper into each team’s story and includes the full audience Q&A.
