Most founders treat AI adoption as a top-down mandate or a series of engineering sprints. John Kim, co-founder and CEO of Sendbird, suggests you might be missing the point – and a lot of quick wins.
Sendbird, a company rapidly shifting to an AI-first mindset, didn't just tell engineers to use AI. They built an internal marketplace, the 'Automator,' where any employee can post a 'quest' for an AI automation. This isn't abstract future-gazing; it's a living system that encourages rapid, bottom-up AI development, complete with experience points and direct executive access for top builders.
Key Takeaways
- Sendbird’s 'Automator' platform creates an internal AI marketplace, letting any employee initiate 'quests' for AI automation tools.
- This system bypasses traditional sprint prioritization, allowing internal AI projects to be picked up by curious colleagues or even AI agents themselves who can read specs and draft code.
- Contributors earn experience points, which convert to real rewards like gift cards, private tea with executives, or company-wide presentation opportunities.
- The platform thrives on engineers' desire for "micro vacations"—small, fun side projects that offer immediate feedback and a quick hit of dopamine.
- Sendbird's approach is best understood through their AI Quest System for Internal Automation framework.
The Sendbird's AI Quest System for Internal Automation
Here’s how Sendbird empowers its team to build AI solutions from the ground up:
- Create a Quest: This is an internal platform where anyone in the company can raise their hand and create what we call the quest.
- AI Agent Building: when there's a quest, AI can actually read through the specification, create PRDS, and start actually coding.
- Internal Learning Guidelines: internal guidelines that continue to get updated pretty much on a daily basis teaching people how to set up GitHubs, create new applications.
- Pre-configured App Templates: app template where all the authentication and all the environments have already been set up. So, all they have to come up is within a cool idea they want to bring to the world.
- Incentives for Builders: people who are completing some of these quests They actually earn experience points. If you earn enough experience points, you can change to a gift card. You can have a tea with any executive you you choose. You can present what you built to the rest of the company.
When This Works (and When It Doesn't)
Kim points out that this system thrives on what he calls "micro vacations." These are moments where builders have some free time and want to work on something not tied to a stressful core product. The quest system provides immediate internal customer pain points, creating a direct feedback loop and an instantaneous dopamine hit when value is delivered. This works best when your company has a strong, curious engineering culture and a backlog of small, frustrating internal inefficiencies that don't warrant full sprint cycles. It also requires leadership trust in giving employees the autonomy to build non-critical tools. It won't work for core product features, or if your team lacks the skills or curiosity for independent problem-solving. If the "quests" are too big, they'll languish; if the company culture is too rigid, no one will engage.
What to Do With This
This week, stop thinking about AI as a separate department. Instead, create a simple internal 'AI Quest' board – even a Slack channel or shared document. Identify one common, annoying internal task that takes 2-5 hours a week from multiple people (e.g., "Automate weekly report generation for client X from HubSpot and Zendesk"). Post it as a quest. Then, take two steps: first, provide a basic "getting started" guide for internal API access and a simple pre-configured app template (a GitHub repo with auth already set up). Second, offer a tangible incentive for completion: a half-day off work, or a $100 gift card. You’ll find a hidden pool of builders eager to solve problems, gain new skills, and earn some quick wins that dopamine hit.