How I AIHow to turn your company into AI builders | John Kim (Sendbird, co-founder and CEO)
CHAPTERS
- 0:00 – 4:09
AI-first as a workforce strategy: turning every team into builders
John and Claire frame Sendbird’s ambition: not just using AI tools, but treating AI as part of the workforce and empowering non-engineers to ship real outcomes. They set up the episode’s central idea—AI adoption should be designed like a product, with enablement and clear expectations.
- •AI-first means embedding AI into how work gets done, not isolated tool usage
- •Empowering marketers/sales/ops unlocks more creativity and faster shipping
- •AI adoption needs structure: tools, training, and internal pathways
- •Outcomes matter beyond “do more with less”
- •Preview of quests, skills marketplace, and token leaderboard
- 4:09 – 5:53
Marketing ships a full swag store in days: “let marketers cook”
John demos a culture-forward swag store built by marketing without engineering support, including payments and playful touches like an Easter egg. The conversation highlights how AI lowers the cost of “fun,” enabling experiences that would never survive a traditional roadmap process.
- •Marketing built a production swag store end-to-end, including Stripe integration
- •AI enables rapid experimentation and customer-delighting experiences
- •Fun features used to be deprioritized because they were costly in engineering time
- •Creative teams gain leverage when they can build and ship directly
- •Example of culture-driven product moments (Easter egg, event promotion)
- 5:53 – 8:18
From backlog battles to side-quest velocity: why the old model failed
Claire and John contrast the “before times,” when marketing ideas had to compete for engineering sprints, with the new reality of cheap iteration. They emphasize the strategic upside: bigger ambition and richer customer experiences, not merely speedups.
- •Traditional sprint planning makes short-lived or playful ideas hard to justify
- •AI makes small experiments economically viable
- •Shifting from ‘middling CMS MVPs’ to bespoke, high-quality experiences
- •AI adoption increases engagement and creative satisfaction across teams
- •Velocity comes from decentralizing building, not just pushing engineering harder
- 8:18 – 9:18
The Automators platform: internal quest marketplace for automation
John introduces Automators, where anyone can create a “quest” and others can collaborate to build automations and tools. It functions like an internal marketplace matching business needs with builders, including visibility into value, risk, and beneficiaries.
- •Any employee can raise a quest (e.g., finance AR/AP workflow automation)
- •Others can volunteer to help, pairing domain experts with builders
- •Completed quests produce reusable assets (repos, skills, demos)
- •Marketplace model avoids heavy prioritization for micro-tools
- •Quest metadata frames impact (risk, time saved, beneficiaries)
- 9:18 – 9:49
AI joins the quest party: agents that read specs, draft PRDs, and start coding
The platform is evolving so AI agents can help build quests by interpreting specs, generating PRDs, and beginning implementation. This sets the stage for hybrid human+AI building and faster internal delivery loops.
- •AI can ingest quest specs and generate structured product artifacts (PRDs)
- •Agents can begin coding alongside human contributors
- •Hybrid collaboration reduces bottlenecks and accelerates iteration
- •Internal documentation updates frequently to reflect fast-moving practices
- •Goal: make building accessible even for non-engineers
- 9:49 – 12:41
Safe-by-default shipping: templates, compliance, and the “happy path” to production
Claire highlights a common failure mode—people prototype locally but can’t ship securely, or ship publicly without guardrails. John explains how Sendbird provides internal templates with auth, environment setup, and compliance built in so teams can focus on ideas, not infrastructure.
- •Internal app templates include auth, environment, and security/compliance baked in
- •Guides teach essentials (GitHub, app setup) to non-traditional builders
- •Reduces risky ‘ship it to Vercel/Netlify publicly’ behavior
- •Standardized stack lowers friction and raises velocity safely
- •Focus shifts to ideation and execution rather than platform plumbing
- 12:41 – 14:56
Gamifying adoption: XP, rewards, demos, and dopamine loops
John describes the incentive layer behind quests—experience points, rewards, and company-wide show-and-tell moments. Weekly showcases reinforce that non-engineering teams can build valuable tools and celebrate progress publicly.
- •Quest completion earns XP that can convert to rewards (gift cards, exec tea)
- •Weekly all-hands demos spotlight what teams built
- •Reinforces social proof: builders aren’t just engineers
- •Creates fast feedback loops and intrinsic motivation
- •Encourages “micro-vacation” side projects that still solve real pain
- 14:56 – 16:06
The AI Engineer for Internal Operations: a cross-functional adoption engine
Sendbird formalizes AI transformation with a dedicated role/team reporting to John and the chief of staff, partnering with CTO and InfoSec. A weekly working group vets tools, unblocks compliance/logging issues, and keeps the internal stack ready for teams to build on.
- •Dedicated AI internal operations engineering role/team drives transformation
- •Reporting line enables cross-functional authority (CEO + chief of staff)
- •Weekly task force with engineering + InfoSec for compliance and tooling decisions
- •Pre-vetted AI stack reduces friction for sales/CS/marketing builders
- •Started bottom-up, then scaled by investing in infrastructure
- 16:06 – 18:56
Company-wide skills marketplace: encoding expertise into reusable plugins
John demos a marketplace where employees publish “skills” and “plugins” (collections of skills) that others can download and reuse. The goal is to prevent duplicate effort and allow teams to co-evolve shared capabilities across functions.
- •Plugins bundle reusable skills; individuals can also share single skills
- •Example: sales plugin using the MEDDIC framework via an advisor skill
- •Marketplace reduces siloed reinvention of similar tools across teams
- •Encourages knowledge capture and repeatable workflows
- •Adoption comes from both executive push and organic pull
- 18:56 – 21:46
Real wins in the wild: marketing’s internal ‘SaaS’ portal and Buzz Board campaign tool
John shares concrete outcomes: marketing built a suite of internal tools (planning, ABM, competitor review) and a Buzz Board that powers a social campaign with metrics and easy posting workflows. These are not generic vendor replacements—they’re custom-fit to how the team operates.
- •Marketing built and runs a full internal tool portal used daily
- •Tools include integrated planning/calendar, ABM, competitor review, and more
- •Buzz Board supports campaign creation, posting, and engagement tracking
- •Supports real-time, culture-specific campaigns (e.g., SF billboard content)
- •Illustrates internal software as a competitive advantage, not a stopgap
- 21:46 – 23:57
“SaaS isn’t dead”—it’s being rebuilt internally (and internal tools finally shine)
Claire and John discuss a shift: teams will increasingly build bespoke micro-software internally rather than buying one-size-fits-all tools. John notes the new era is uniquely fun for internal tooling—better UX, faster iteration, and fewer resource constraints with AI.
- •Internal tools become viable, polished, and fast with AI assistance
- •Organizations will build custom-fit workflows before seeking external vendors
- •Value comes from cultural/workflow alignment, not feature parity with SaaS
- •Internal tools teams move from under-resourced to high-leverage
- •AI improves responsiveness, design quality, and iteration speed
- 23:57 – 30:57
Token dashboard and adoption measurement: tiers from newbie to “AI god”
John demos token tracking at company, team, and individual levels, with tier labels to guide enablement. They stress measurement as expectation-setting—not punishment—and use the data to identify where teams are in the adoption journey and how to level up.
- •Company-wide token usage tracking with team and manager views
- •Tier system (beginner → intermediate → expert → architect → catalyst → AI god)
- •Used for coaching conversations, not performance reviews
- •Model/tool choice patterns emerge (Claude Code vs Codex by work type)
- •Goal includes smoothing usage curves to enable around-the-clock AI “partners”
- 30:57 – 35:22
Personal AI workflows: knowledge gardening and building private learning worlds
John shares personal use cases: an open-source “Gardener” that enriches and organizes notes, and a method for generating offline, structured learning hubs on complex topics. Claire reflects on AI as a force for deeper learning rather than cognitive decline.
- •“Gardener” project enriches notes: entities, research, cross-links, cleanup
- •AI-generated learning centers create structured maps of a subject in minutes
- •Examples: neuroscience, quantum mechanics, fusion, startup research clusters
- •Offline, personalized knowledge bases outperform generic web experiences
- •AI can increase depth of engagement by adapting to individual learning styles
- 35:22 – 42:19
Leadership playbook and lightning round: champions, signaling, games, and being nice to AI
In rapid-fire Q&A, John gives a practical adoption recipe: find internal champions, celebrate them, and ensure leadership models the behavior through high usage. The conversation closes with personal anecdotes—pro gaming background, builder energy, and a humorous take on treating AI politely.
- •Find high-curiosity/high-agency champions and amplify their stories
- •Adopt a “fail forward” culture to reduce fear and build momentum
- •Leaders must model usage—top token consumers include CTO/co-founder
- •Hiring shifts toward curiosity, agency, and energy over tenure
- •Lightning topics: gaming influence, prompting style, and polite AI collaboration
