I recently came across this article, 25 proven tactics to accelerate AI adoption at your company, and loved it! It’s simple to follow and remember, and has tons of examples.

It got me thinking － could we do a similar “guide” for L&D professionals trying to adopt AI in their teams? The purpose would be simple, cut through the noise of “what you should do” and just share what people are actually doing, and what’s working for them.

I naturally turned to my network, because they’re awesome, very into AI, and always open to sharing their practices. So what you see below, it’s 100% thanks to them, their courage to experiment, and openness to share.

I’ve asked L&D leaders from 9 companies the same questions, and here are the 5 red threads I found in their stories about AI adoption within their teams:

1. Start with your biggest pain points, not the flashiest tools
2. Create psychological safety for experimentation
3. Make friends with IT, Legal, and Security early
4. Turn sharing into a systematic practice
5. Measure adoption, not just outputs

Let's break down exactly how these L&D teams made it work.

1. Start with your biggest pain points, not the flashiest tools

Rather than chasing the latest AI tools, several L&D teams found more success by identifying what was eating up the most time and energy in their workflows, then finding AI solutions that could address those specific friction points.

Here's what this looks like in practice:

Focus on high-volume, repetitive work first: Marie at Witherslack Group identified content creation as their perfect storm challenge, "not being able to increase our resource, large volume of learning requirements, increasing geographical spread of learners." They started with Synthesia for quick video creation, creating "something in about 20 minutes that looked polished and professional."

Pick one friction point per person: Christine at BDO USA asked each team member to "run a small, targeted experiment on a real deliverable. Pick one friction point, use AI to remove it, and evaluate effectiveness with a human review." This personalized approach meant everyone found something immediately useful.

Address your customers' real needs: Amanda at e-Core didn't just build content faster, she listened when leadership pipeline participants said "they love when we curate content, but they don't have time to read the materials, and podcasts would be a better fit." She used NotebookLM to turn existing materials into podcasts that fit their actual consumption preferences.

Start with your most time-consuming processes: Chris at Booking.com focused first on "automation within our learning operations" because "we had poor data and lots of manual processes." Rather than jumping to content creation, they addressed foundational workflow issues that were slowing everything else down.

The key insight: successful AI adoption in L&D isn't about using the coolest tools. It's about solving real problems that are already slowing your team down or preventing you from serving your learners better.

2. Create psychological safety for experimentation

Several teams emphasized creating an environment where team members felt safe to experiment, fail, and share what they learned, without judgment or pressure to immediately show ROI.

Here's how they built this safety:

Establish "no wrong answers" spaces: Christine created "a dedicated, psychologically safe place for all interested team members to discuss their thoughts about AI and share their usage stories." Within clear guardrails, "there were no wrong answers" and they could "change directions several times, and shared 'failures', which I still consider wins."

Role model curiosity over expertise: Faye at AELIA emphasized "constant sharing of practices and results as well of prompts, while we also comment on the prompts." She's "persistent in explaining the 'what' and the 'why'" and encourages others to share, even when they're not comfortable yet.

Give people permission to invest time: Emily at CreateFuture's organization "recently released an AI experiment budget, encouraging people to find, adopt and share new AI tools" with "a robust and clear experiment policy" that "makes it less stressful for them to try new things."

Start without a rigid strategy: Eman at talabat describes their initial approach: "There was no strategic approach at the beginning, it was more about wanting to try different things and seeing what worked, or based on what we were working on we used to think of how can AI support us here." This organic exploration helped build confidence before adding more structure.

The bottom line: teams that succeed with AI create cultures where experimentation is expected and "failures" are treated as valuable data points, not performance issues.

3. Make friends with IT, Legal, and Security early

Almost every L&D leader mentioned governance and approval challenges. The teams that moved fastest were those who proactively built relationships with their IT, Legal, and Security teams rather than asking for forgiveness later.

Here's their approach:

Get ahead of the approval process: Marie at Witherslack Group "built relationships with our Data Protection and IT Teams to let them know our plans to experiment/play and to identify the process for gaining approval." This meant when team members got excited about a tool, there weren't unexpected IT blocks.

Start with low-risk experiments: Adam at Service Express prioritized tools "by impact × feasibility × risk, with a quick data sensitivity check with IT/Legal." Their People Pal chatbot was built on "approved SharePoint content with clear guardrails and human review", making it easier to get buy-in.

Assign someone to own the relationship: Some teams went further by "assigning a lead to own working with procurement, legal, and engineering to fast-track AI tool approvals and eliminate bottlenecks," similar to what Zapier did in the original article.

Be transparent about what you're doing: Rather than experimenting in secret, successful teams shared their plans upfront. As Marie noted, "What we are doing is low risk data and security-wise, which helps, but there is still some nervousness at times from an IT Security and data perspective."

The key: treat IT, Legal, and Security as partners in innovation rather than obstacles to work around. They want to enable the business, they just need to understand what you're trying to do and why.

4. Turn sharing into a systematic practice

Rather than relying on ad-hoc sharing of AI wins, several teams created systematic ways for team members to learn from each other's experiments and build on each other's discoveries.

Here's how they systematized sharing:

Create dedicated channels for AI sharing: Heather at Real Chemistry "shares everything, like EVERYTHING." When she discovers something new, like "a Canva webinar about how to translate videos into other languages," she immediately shares it with the team. Some teams created specific Slack channels for AI tips and workflows.

Build regular sharing rhythms: Emily at CreateFuture made "time in our L&D and wider People team meetings to share practical use cases and celebrate small wins. This approach makes our AI adoption feel collaborative and low-risk, not like a top-down mandate."

Establish communities of practice: Christine runs "a monthly community of practice to share research and results" where the team discusses both what's working and what isn't. Faye describes trying to establish a "Community of Practice" where she keeps "sharing my prompts and the results produced as well as all my new discoveries of tools."

Model experimentation openly: Eman at talabat describes how "we role modelled the use of AI and sometimes when approaching tasks and projects we would ask, how can AI help us here." They also "openly communicated the importance of the team exploring AI and we were open to approving budgets for different tools."

Successful AI adoption happens when learning compounds across the team. Individual experiments become team capabilities when there are systematic ways to capture and share what works.

5. Measure adoption, not just outputs

While many L&D teams measure learning completion rates and satisfaction scores, some teams also track different metrics, focusing on adoption behaviors and process improvements rather than just content creation speed.

Here's what they measure:

Track experimentation and usage patterns: Adam at Service Express tracks "adoption, time saved, and deflected queries" for their AI tools. They also monitor "top queries" from their People Pal to identify content gaps and inform their L&D roadmap.

Measure process improvements: Christine notes they "showed up as shorter project cycle time" from their early AI experiments. Rather than just counting how much content they created, they measured how AI changed their workflow efficiency.

Monitor learning and skill development: Several teams track how many people are actively experimenting with AI tools, sharing learnings, or building new capabilities. Emily mentions their organization tracking people's use of the AI experiment budget as a proxy for adoption.

Document what changes and why: Christine wishes she had "defined an evaluation strategy from day one" to systematically capture "How many experiments have we tried? How many workflows are more automated? What changed and why?"

The insight: measuring AI adoption requires different metrics than traditional L&D measurement. Focus on behavior change, process improvement, and capability building rather than just output volume.

Where to go from here

The L&D teams succeeding with AI understand that adoption isn't about implementing every new tool that comes out. Instead, they start with real problems, create safe spaces for experimentation, build partnerships with governance teams, systematize learning sharing, and measure the behaviors that drive real change.

The goal is to empower every L&D team member to work with AI to solve problems better and faster. As these interviews show, someone on your team might discover an AI workflow that transforms not just your department, but how your entire organization learns and grows.