Every organisation is chasing AI Fluency as a key employee capability, but very few are succeeding beyond the vanity metrics.
If every company has decided that AI will be the infrastructure of everything it does, making fluency stick is so much more than a one-off course or “learning week”.
Here’s a roundup of 13 lessons from AI enablement leaders at Stripe and Canva to make sure your efforts don’t end in the graveyard.
Building AI Fluency: A L&D Practitioner’s Playbook
Let’s start with a practitioner-led conversation on driving meaningful AI adoption in organisations from someone who has done the work.
Here’s my biggest takeaways from a conversation with Galen Crawford, the human who led AI enablement from the L&D side at Chime and Stripe.
1. Fluency is making AI mean something in your actual day-to-day work
Galen recommends you start with the boring, repetitive, low-stakes tasks, then build up to pulling apart your own messy 12-step process to find the few steps where AI delivers meaningful ROI.
Most organisations need intermediate-level fluency as the target, not advanced. Specific builder roles (forward deployed engineers, AI workflow operators) sit above doing things like turning an FAQ into an AI workflow or building a benefits Q&A bot.
2. Don’t buy the hype
Most organisations are two years behind whatever you’re seeing online, and most of it is embellishment.
This convinces people everyone else has already cracked this, so they always feel behind. The AI bros and their slick workflows aren’t the reality. That stuff just doesn’t land inside a real enterprise.
Galen quite rightly pointed out that “hype has become the enemy of AI adoption”.
3. Share your failures
I’d love more companies to do this but I’m also a realist.
I know they don’t want to.
Galen told us at Chime they’d openly share the prompts and AI projects that flopped, right down to what the failure cost them in tokens, then treat it as a brilliant example of what not to do.
You’ll never build fluency in a culture where one failure means “bin the whole AI thing.”
Instead, Galen encouraged us to create ‘low-stakes environments’ where people can experiment, fail, and learn. When proof of concept is achieved and low-stakes environments exist, you can then watch tools scale naturally: experiment → personal use → a colleague finds it useful → organisational scale.
4. Build a proof of concept
I find that most of the time, you need to show people ‘the enormity of the possible’ because very few get there organically.
A good example of this is when Galen built a custom GPT for the performance review process. 1,500 people got value from it with around 10,000 chats running through it.
Then he handed over the master prompt for folks in other teams to use as an example of ‘what good looks like’ to take and rebuild for their own use cases.
5. Stop “doing elearning” and start practising with AI
You already know I hate elearning, so I was totally sold on everything Galen shared with running live, in-tool sessions where 80% of the time is people solving real problems with AI.
So, yes, use AI to learn AI, and please, stop reporting completion and usage rates like they’re a success. Someone logging into Copilot tells you nothing about whether they’re any good with it.
6. Start treating AI fluency like any other business capability
We gotta embed it in our processes.
Being genuinely AI-native yourself (you can’t teach what you don’t practise) is critical to build trust and credibility as a L&D partner. Don’t forget to make it useful and measure impact at scale rather than vanity metrics like number of chats.
7. Build your conditions for success
While I’ve left this as the final point of my conversation with Galen, it was one of the most important.
“Some people at senior management level understand the importance of driving AI fluency from an L&D or enablement function but without executive (C-suite) alignment and buy-in, programmes end up dead on arrival and easy to put on the chopping block. Push to get it in front of the C-suite for at least a nod.
Rather than going through success metrics and needs analysis with them, create a simple “conditions for success” slide:
- Here’s what you want from this
- Here’s what we need from you: visible buy-in, understanding that these things take time, consistent application
If leadership later asks why people aren’t orchestrating agents and making millions, you can point back to the conditions for success and show which ones weren’t met.”
A huge thank you again to Galen for sharing these insights.
Six Insights from Canva’s AI week that totally changed how they approached AI enablement
I gotta be honest, I don’t read Fortune often.
The attack of the ads usually forces me to retreat before I finish reading anything (sorry, Fortune), but I’m so glad I got through this one.
Canva’s Chief Customer Officer, Rob Giglio shared what really happened when they gave the entire Canva team a dedicated week to experiment with AI.
This was incredibly refreshing to read in a sea of people pounding their chests about “AI success”.
Here’s the insights that stuck with me:
1️⃣ “Deploying isn’t the same as enabling”
Just because they have access to ‘x’ AI tools doesn’t mean people know how to leverage it meaningfully.
We know this well with LXPs and the LMS.
Most sit dormant across many organisations. I’ve worked with very few organisations where teams knew how to use their local learning platform or that it even existed at all.
So, while your org might have an AI tool license for everyone, it doesn’t mean they’re using it to impact performance.
2️⃣ Enablement “doesn’t happen in a single lunch-and-learn”
Sorry to disappoint you but that one-off learning at work week session won’t cut it.
You’re smart, so I know you know this already.
There is no such thing as “Teaching AI”. Every business and every team have different requirements and maturity levels. That leads me to our next insight.
3️⃣ Design programs to “meet people where they are”
Generalised one size fits all approaches won’t get you far. You need to layer your approach based on roles, tasks and skills.
For Canva’s go-to-market teams (GTM) this involved getting to grips with their internal AI tool stack, creating dedicated space for play and build sessions, and layering in role specific workshops from partners like Anthropic, OpenAI, and Google.
4️⃣ “A learning week is a catalyst, but what comes after is the real test.”
Rob explained they could have just stopped with the AI week but that wouldn’t have had much of an impact. To move the dial, they built long-term infrastructure.
Canva deployed:
- AI Hub with self-paced courses, toolkits, and templates
- Fortnightly AI Forums to surface practical use cases from across the business
- A network of AI Exemplars who lead regular roadshows on emerging tools and breakthroughs from inside the company.
- AI Show & Tell where product and research teams showcase the latest developments.
And Rob was quick to point out to other leaders: “You don’t need to copy our model, but you do need to make AI adoption part of your culture, not just mandatory training.”
5️⃣ “The challenge is helping teams build enough confidence and fluency to actually change how they work.”
As I keep coming back to, tools are only one part of the equation.
Rewiring how we work and learn is something else.
We both know this won’t be achieved in any AI week. Much like Canva discovered, you must go beyond access and deal with the more difficult stuff of behaviour change.
I’ve always found using tools, and getting people to use tools easy.
But getting them to unbundle their mindset from what they know to where they need to be is really tough. You’re dealing with ingrained habits and beliefs that won’t change overnight.
6️⃣ “People didn’t know how to give themselves permission to experiment.”
As Rob highlights in the article: “We hadn’t built the conditions for genuine exploration.” and so the teams “Defaulted to the use cases they already knew rather than exploring ones that might change how they worked”.
While access is great, without the right environment, don’t expect a miraculous transformation.
This to me was probably the most interesting insight of the whole article for L&D pros. We hear so much from our organisations that they don’t have enough time to learn. Most orgs aren’t going to set aside time for everyone to do that, yet Canva did and it didn’t go the way they expected.
People overestimate what they can do in a week, but underestimate what can be achieved in a year.
Of course, social media makes us feel like we’re all behind. Yet, Galen reminded us in the first part of today’s conversation that most companies are not at a level of what you see on social media and that’s totally fine.
Even if we give people protected ‘time to learn’, it doesn’t mean they actually do. That’s why Rob’s earlier point on making AI fluency part of your culture is far more powerful than one-off events on a topic.
Final Thoughts
Ok, folks. There we have it.
Lessons from the frontline of AI enablement with an L&D focus.
Plenty for you to add to your toolkit and make AI enablement part of your culture. I’d love to know any of these that stuck out for you, and anything you think is missing.
Before you go… 👋
If you like my writing and think “Hey, I’d like to hear more of what this guy has to say” then you’re in luck.
You can join me every Tuesday morning for more tools, templates and insights for the modern L&D pro in my weekly newsletter.
