The AI Adoption Playbook Big Companies Are Stealing From — And What Solo Operators Can Steal Back
John Kim's playbook at Delight.ai uses quests, token leaderboards, and a skills marketplace to push AI adoption to the whole company. Here's what's actually portable to a small team.
I just read Lenny Rachitsky's interview with John Kim, the CEO of Delight.ai (formerly Sendbird), and the playbook he laid out for getting AI into every part of his company is one of the most thought-out frameworks I've seen.
Most companies are still in the "announce ChatGPT licenses, hope for the best" phase of AI adoption. Kim's team built quests, token leaderboards, an Automators platform, and a skills marketplace. Real adoption, not just access.
Let me walk through the parts that I think actually translate down — to a marketing agency, a small SaaS team, even a solo operator like me.
What Kim's playbook looks like inside Delight.ai
- Automators platform. Any employee can request an AI tool. Engineers build it. Marketing built a fully working e-commerce swag store with Stripe integration in days, no engineering needed. Sales built their own CRM tools.
- Quest system. Gamified projects with tracked progress and visible wins.
- Token leaderboards. A five-tier dashboard from beginner to "AI God," measuring usage curve, not raw consumption.
- Skills marketplace. Internal directory of every AI capability the company has built.
- Visible leadership usage. Executives use AI publicly. That's the most powerful adoption signal in any company.
- Secure templates. Pre-built compliant frameworks so non-technical teams can ship to production safely.
What I love about this is that none of it is about "having AI" — it's about what's blocking adoption in real companies. The blocker is rarely the tool. It's friction, fear, and the absence of a path.
What's portable for the rest of us
Most of my clients aren't running a 500-person company. They're running an agency, a coaching business, a SaaS, a clinic. So which pieces of Kim's playbook actually scale down?
1. The Automators idea, in micro-form. I've seen this in my own work — when one person on the team learns to build a small AI workflow and publishes it for everyone, adoption explodes. Even if you're a team of three, having a shared Notion page of "prompts and Claude projects we've built" is the entire skills marketplace.
2. Visible leadership usage. This one matters whether you have 5 employees or 50. The team watches you. If the founder is using AI on calls, in emails, in research — the team will. If you don't, they won't.
I've been in calls with business owners who say "my team doesn't get AI yet," and I usually ask: do you get AI yet? Show them. Use it on the call. Live.
3. Measurement without punishment. Kim explicitly designed the leaderboard to track adoption curve, not police consumption. That's a leadership move, not a tech move. The fastest way to kill AI adoption is to grade your team on whether they used it "correctly."
Where this connects to my own experience
I've spent a chunk of the last year watching small business owners try to bolt AI onto operations they already had. The pattern is always the same:
- They buy ChatGPT seats or Claude subscriptions.
- A few people use them.
- Most don't.
- A year later, they cancel half the seats and decide "AI didn't work for us."
It didn't work because there was no path. No quest. No visible win. No template. No leader using it. Kim's playbook is interesting because every piece of it is designed to remove a real friction. Most companies skip the friction work and wonder why nobody adopts.
If you're a founder reading this — even with a team of two — your job is the friction. Build the templates. Show the wins. Run AI in front of your team. The rest follows.
What I'd do tomorrow
If I had to install one thing from Kim's playbook into a small business this week, it would be the Skills Marketplace in its simplest form. A shared doc. Every prompt that worked. Every Claude project somebody built. Every workflow that saved an hour.
That one doc, kept current, does 60% of what Kim's marketplace does — for a team of any size.
The Bottom Line
Elite AI adoption isn't about having access to the best tools. It's about removing the friction between an employee and the tool. Kim's playbook at Delight.ai is the cleanest example I've seen of that — and most of it is portable down to a team of two. Pick one piece and install it this week.