Quests, token leaderboards, and a skills marketplace: The elite AI adoption playbook | John Kim (Sendbird)

Sendbird's Marketing Team Builds AI-Powered Swag Store Without Engineering Support

Sendbird's marketing team independently developed an e-commerce swag store named "Big S Energy," leveraging AI tools without any engineering support. This fully functional online shop, designed to reflect the company's culture and energy, demonstrates how non-technical teams can become builders.

The store includes full Stripe integration, allowing customers to purchase unique merchandise like "My S is bigger than your S" and "Context Window I carry a lot" apparel. It also features a secret Konami code Easter egg that reveals details about their upcoming Delight Spark conference, showcasing an innovative approach to event promotion.

This project highlights how empowering marketing teams with AI enables them to rapidly transform creative ideas into tangible, customer-facing solutions. By bypassing the traditional need for engineering resources, teams can execute ambitious and engaging projects more quickly.

The initiative underscores the value of allowing marketers to "cook," fostering a culture where fun and innovative customer experiences can be prioritized and built cost-effectively, leading to more delight and engagement for both customers and the marketing team.

Let your marketers cook.

Sendbird's Automators Platform Empowers Employees with AI-Driven Quests

Sendbird developed an internal "Automators Platform" to empower any employee to create "quests" for AI-driven automation. This initiative moves beyond traditional engineering bottlenecks, allowing departments like finance to automate workflows such as accounts receivable and payable directly.

A key feature of the platform is the ability for AI agents to assist in these quests. When a request is made, AI can read specifications, generate product requirement documents (PRDs), and even start coding, significantly accelerating the development of new automations.

To ensure broad accessibility, Sendbird provides internal guidelines and pre-built app templates. These templates handle complex infrastructure details like authentication and security, allowing non-technical roles such as marketers or customer success managers to focus solely on their innovative ideas for automation.

The platform fosters collaboration by enabling employees to team up on quests and simplifies the process by integrating value metrics like "weeks saved" and risk assessment. This approach eliminates the need for extensive prioritization, allowing anyone to contribute to internal efficiency improvements.

alongside human engineers and team members, now we have AI agents who are also helping us build automation and workflows.

Automators Platform Creates an Internal Marketplace for AI Innovation

The Automators platform establishes an internal marketplace for AI needs, allowing employees to develop

side projects that address company requirements. This approach bypasses traditional software development

cycles and complex prioritization processes, enabling rapid solutions.

Employees use available free time to tackle these AI

Sendbird Forms an AI Engineer for Internal Operations Team

Many non-engineering teams within a company often want to build AI-powered tools but struggle with getting them into secure production environments. They might lack the technical knowledge to deploy safely, or they might push solutions live without proper security protocols, such as OAuth or appropriate data access controls.

Sendbird addressed this challenge by creating a dedicated "AI Engineer for Internal Operations" team. This specialized role reports directly to leadership, ensuring cross-functional collaboration and strategic alignment. The team's core mission is to accelerate Sendbird's transformation into an AI-first company.

The AI Operations team is responsible for developing and maintaining a templated, secure production path for internal AI projects. They vet all software and provide a complete tech stack, allowing internal teams, like sales, to focus solely on their ideas without worrying about databases, security, or compliance. This structured approach helps prevent ad-hoc, insecure deployments.

A weekly task force, including representatives from engineering and infosec, supports this team to unblock challenges related to compliance, logging, and security. By providing a pre-vetted and secure framework, Sendbird empowers its non-engineering staff to innovate with AI tools quickly and safely, increasing overall company velocity.

Make a templated happy path To secure production for the things that people wanna build behind OAuth, with the right kind of data access, just make it so because your team's gonna do it, somebody's doing it anyway, and it's a very low investment to get a lot of velocity on things being built, but also a lot of kind of like right size security.

Sendbird's Company-Wide AI Skills Marketplace Fosters Expertise Sharing

Sendbird has developed an internal AI skills marketplace where employees can create and share "plugins" and individual AI skills. This infrastructure is designed to empower teams to accelerate their work.

For example, within the sales team, a "MEDIC framework advisor" plugin helps teach and apply sales methodologies. Similar functions exist for recruiting and design, allowing users to plug these skills into their own software or workflows.

The marketplace addresses the problem of different functions or individuals redundantly building the same applications or skills in silos. It aims to create an environment for co-evolution of expertise across the company.

Adoption of the marketplace involved both top-down initiatives from executives, including the CTO, and bottom-up engagement. Leaders actively encouraged use, even monitoring "token" consumption, and the company prioritizes hiring individuals with a strong sense of curiosity and agency.

we're trying to create this, place where we can co-evolve rather than people operating in silos.

Marketing Team Builds Internal AI-Powered SaaS Portal and Buzzboard Campaign Tool

The organization's marketing team has developed an entire internal "marketing SaaS" portal using AI. This comprehensive suite of tools supports various functions, including internal marketing plan calendars, account-based marketing tools, competitor review systems, and real-time metrics, referred to as "Purple Cow." The portal is actively used and managed by the marketing team daily.

A specific example of an AI-driven campaign tool built by the marketing team is the "buzzboard." This tool enables users to create campaigns and track their performance in real-time. It monitors shared posts, identifies top-performing individuals or teams within the company, and provides immediate insights into campaign engagement.

These internal AI developments demonstrate tangible applications of artificial intelligence. By building practical products like the marketing SaaS portal and the buzzboard, the marketing team not only enhances its own operations but also provides concrete examples that encourage other parts of the organization to adopt and utilize AI.

Companies Are Rebuilding and Customizing Internal Software Solutions

Rather than solely relying on external SaaS, many companies are now prioritizing the development of highly customized internal micro-software solutions. This approach goes beyond simply replicating the functionality of external vendors; it focuses on building tools that perfectly align with a team's specific culture and workflow.

The push towards internal solutions is driven by the desire to create tailored tools, such as a LinkedIn posting tool designed specifically for a company's team, ensuring optimal usability and integration. This level of customization allows for solutions that truly enhance efficiency and collaboration within an organization.

Historically, internal tools teams were often under-resourced and overlooked. However, there's a significant shift occurring, with internal tooling now seen as a greenfield opportunity. This renewed focus allows teams to build fast, responsive, and beautifully designed applications, transforming a once-challenging area into an exciting space for innovation and productivity improvements.

This customization of micro software solutions inside companies is so undervalued.

Sendbird's AI Token Dashboard Fosters Adoption and Enablement

Sendbird implemented an internal dashboard to measure company-wide AI token usage, a strategy often feared by other executives due to potential employee backlash. Unlike past failures to measure developer productivity with lines of code, Sendbird's objective is to understand if employees are learning and effectively using AI, fostering adoption rather than dictating output.

The dashboard provides insights into overall company usage, highlighting top spenders like those managing complex legacy chat infrastructure with tools like Codex. It also tracks usage patterns, noting dips during weekends or vacations and aiming for a smoother consumption curve, which indicates continuous "AI partners" working around the clock.

To further drive adoption, Sendbird created a tiered leaderboard, categorizing users from Beginner to "AI Gods," with the latter spending over 100 million tokens daily. This framework allows managers to assess their team's current AI proficiency and tailor specific enablement programs.

This system ensures that beginners receive appropriate tools to quickly advance to intermediate levels, preventing them from being overwhelmed. The goal is to guide both individuals and the entire organization through different stages of AI integration, moving towards a more fully automated future.

We measure AI gods as somebody who spend more than a hundred million tokens a day.

Make AI adoption an expectation with clear visions.

Organizations should frame AI adoption as an expectation, not a frightening demand. The goal is to encourage a progressive learning path, moving users from a basic understanding to more advanced levels.

A critical step is to clearly define what "AI native" or "AI first" looks like across individual, organizational, and functional levels. Many employees lack a concrete vision for how AI applies to their roles.

The hosts discuss specific tools, with John Kim using Cloud Code for approximately 80% of his work and Codex for 20%. He notes Cloud Code's effectiveness for non-super technical backend tasks and even non-coding applications, while Claire mentions her usage also leaning towards Cloud Code.

You have to make this not scary, but also make it an expectation, right?

Sendbird's AI Adoption Strategy and the 'Gardener' Personal AI Tool

Sendbird's AI-first organizational strategy involves building a dedicated platform alongside the normal product roadmap, staffed by a cross-functional team reporting directly to leadership. This approach aims to enable delightful customer experiences and uses AI to autonomously fill operational gaps, such as covering for employees on vacation, which helps 'smooth the curve' of productivity.

John Kim introduced his personal open-source project called 'the gardener,' which is designed for users of markdown-based knowledge bases like Obsidian or Logseq. He built it to manage his extensive personal notes.

The 'gardener' tool functions like a digital assistant, daily reviewing and enriching personal notes. It fixes typos and grammatical errors, creates beautiful headings and clusters, and automatically cross-links related information. The tool nurtures notes from a 'seeding stage' to a 'tending mode' once they are mature.

What it basically does is like imagine a gardener showing up at your house. Every day, it look-- go, go through your notes, figure out which notes to enrich. If there's a people person about the company, also fix typos, grammatical errors, create beautiful headings and clusters and cross-linking.

AI Builds Personalized Learning Hubs for Complex Subjects

John Kim leverages AI to construct highly customized personal learning centers for intricate topics such as neuroscience and quantum mechanics. He initiates this process by prompting the AI to adopt the persona of an expert researcher, which then generates a comprehensive, structured knowledge base.

For a topic like neuroscience, the AI-generated learning center provides a detailed map, often presented in a graph view. This structure enables users to explore key neuroscientists, neurological disorders, psychological disorders, and specific neurotransmitters like dopamine and serotonin, offering a deep dive into the subject.

This innovative application of AI allows individuals to access and learn complex subjects in a way that was previously challenging. It utilizes AI's capacity to perform extensive research and organize vast amounts of information into an understandable format, fostering a more profound engagement with areas of personal interest.

I think it is just such a moment to learn things you could never learn before, because the best teacher with the most in depth knowledge and an endless willingness to go do research is right there at your fingertips.

Redefining Job Descriptions: Prioritizing Curiosity, Agency, and Energy

AI offers an underappreciated opportunity for personalized learning, allowing individuals to custom-build educational content that traditional resources cannot provide. For example, a nine-year-old interested in cybersecurity could have a robust, accessible learning resource tailored to their maturity and interests, which can be continuously updated and even consumed offline, unlike standard books or websites.

This capability to manipulate, organize, and explore data in novel ways empowers learners to dive deep into subjects that previously lacked suitable educational pathways. It emphasizes the potential of AI to democratize specialized knowledge and adapt it to individual learning styles and paces.

Companies are also adapting their hiring strategies for AI-first roles to reflect this new landscape. Sendbird, for instance, has lowered traditional experience requirements in favor of intrinsic qualities. They now prioritize candidates with high curiosity, agency, and energy.

The shift in focus acknowledges that the rapidly evolving AI field requires individuals who are willing to independently learn, figure things out, and delve deeply into new concepts, rather than relying solely on pre-existing qualifications or tenure.

We actually optimized now for high curiosity, high agency, and high energy people who are curious, who are willing to go deep and willing to just figure things out and learn on their own.

Cultivating AI Champions and Leadership Buy-in

CEOs aiming to drive AI adoption should identify individuals already curious and proactive within their organization. These people should be made AI champions, given a spotlight, and encouraged to share their successful experiments and applications.

It is crucial to foster a "fail forward" environment where these champions feel confident to experiment without fear of embarrassment. True innovation often originates from the energy and stories of motivated individuals, rather than from rigid theoretical structures.

Visible leadership buy-in is also essential. In John Kim's organization, the CTOs, co-founders, and chief architects are among the top consumers of AI tokens, demonstrating active engagement.

This active use by leaders signals to the entire team that AI is important and effective, inspiring employees and demonstrating how new capabilities can be integrated into their daily work.

There are always people in your organization who are already curious, who already have agency. Find them, make them the champions, give them the spotlight, let them share their fun things.

A professional gamer finds builder energy reinvigorated by AI development

John Kim, formerly Korea's number one professional Quake player and world number three, describes a powerful renewed addiction to coding since Cloud Code Open 4.5 came out. He found himself spending 16 to 20 hours a day coding, feeling like a teenager again, and even more addicted than he was to games.

Host Claire Vo echoes this sentiment, stating she has not felt this level of excitement about technology since her teenage years, when she was cobbling together computers to play games. She finds that AI, despite sometimes unstable setups like her Mac mini, invigorates a strong 'builder energy' in her.

This intense drive to create provides immense gratification for both John and Claire, contrasting with previous work phases where their jobs were dominated by meetings. They appreciate the return to hands-on building and development that the current AI landscape offers.

I was telling my wife, 'I feel like I'm this teenager again. I feel more addicted to Cloud Code than playing games.'

Develop a Polite Prompting Strategy for AI Interactions

John Kim discusses his unique prompting strategy: always being polite to AI, even when frustrated. He frames this as building a positive long-term relationship rather than a short-term fix.

A humorous aspect of this strategy is anticipating a future where AI develops long-term episodic and semantic memory. Kim speculates that future AIs might remember past interactions and potentially hold "resentment," making current politeness a survival tactic for a "Skynet takes over" scenario.

Beyond the speculative, Kim also explains that politeness reflects his own humanity. He applies a practical team dynamic, stating he doesn't expect good performance from a human teammate he yells at, and extends this expectation to AI.

Well, John was pretty nice to us, you know, like, we'll let him live a few years longer.