Key Takeaways

  • Track AI usage, but don't tie it to performance: Sendbird actively measures company-wide AI token consumption, not for employee review, but to guide and enable AI adoption across teams. This nuance is critical for buy-in.
  • Segment users for targeted enablement: Sendbird classifies employees into 'AI God' tiers (Beginner to AI God, with specific token consumption thresholds like 100M tokens/day) to help managers understand skill levels and tailor support. No more one-size-fits-all training.
  • Identify your organization's AI tooling preference: The consumption dashboard highlighted that while Sendbird is a "Claude Code shop" overall, their top AI spenders actually favored Codex, revealing a deeper trend in expert usage.
  • Aim for continuous AI work, not just sporadic bursts: John Kim emphasizes "smoothing the curve" on token consumption. A steady, consistent usage graph means AI partners are working around the clock, even when humans are offline, moving towards autonomous work.
  • Use Sendbird's AI God Tiers for Token Consumption as a progression roadmap: This framework helps define, measure, and encourage the journey from novice to master in AI tool utilization.

The Sendbird's AI God Tiers for Token Consumption

Sendbird uses these tiers to categorize employees based on their AI token consumption, providing a clear path for progression and targeted enablement:

  • Beginner: Beginner level of AI token consumption.
  • Intermediate: Intermediate level of AI token consumption.
  • Experts: Expert level of AI token consumption.
  • Architects: Architect level of AI token consumption.
  • Catalyst: Catalyst level of AI token consumption.
  • AI Gods: Somebody who spend more than 100 million tokens a day.

When This Works (and When It Doesn't)

This framework shines when your goal is to deliberately level up your team's AI skills, not just monitor activity. As John Kim puts it, “knowing where your team is on the journey then you can tailor what kind of enablement you want to do for them.” It encourages an organizational journey, from basic automation to fully automated AI processes. By accepting that it's okay to be a beginner, managers can give precise tools to move employees to the intermediate stage, rather than overwhelming them with advanced concepts.

However, this framework has limits. It works best in environments where AI token consumption is easily trackable and directly correlates with skill progression. For early-stage startups with tight budgets, 100 million tokens a day might be an unreachable benchmark, making the 'AI God' tier feel aspirational rather than attainable. The focus on token count might also overlook qualitative AI skill—someone using fewer tokens more strategically could be more effective than a high-volume user. It also requires a culture of trust; if employees fear the leaderboard will be misused for performance reviews, it can backfire into reduced, not increased, adoption.

What to Do With This

Forget the generic "experiment with AI" advice. If you're running a lean startup or leading a small team, apply Sendbird's thinking immediately. First, adapt the "AI God" tiers to your own context. For instance:

  • Beginner: Uses ChatGPT for brainstorming, basic drafting.
  • Intermediate: Integrates specific AI tools (e.g., Midjourney for design, Bard for quick research, an AI code assistant for refactoring).
  • Expert: Automates repeatable tasks with custom AI scripts or integrates AI APIs into existing workflows.
  • Architect: Designs end-to-end AI-powered processes for entire business functions (e.g., marketing content generation, customer support triage).
  • AI God: Your internal benchmark for peak autonomous AI operation—perhaps a specific AI agent built that runs continuously, handling a key business process without human intervention.

Next, implement a simple, low-overhead way to track your team's (or even your own) AI usage. This could be a shared doc where team members log major AI-assisted tasks, or even just a bi-weekly check-in on "what AI tools did you use this week, and how?" Use this data to identify where your team falls on your adapted tiers. If most are Beginners, focus on targeted, specific tutorials for their roles. If you have an Expert, challenge them to build an AI solution for a company pain point. Finally, ask yourself: what processes can you make so "smooth" that your "AI partners" are working around the clock, even when you're not?"

not? That's your path to true AI leverage."