Alem TuzlakAI is being bolted onto every product right now, and most of the results are throwaway chatbots that frustrate users. There are great ones out there, like Prisma's built-in chatbot that allows you to talk to their docs, and then there's ones that have no guardrails and that act like a proxy to ChatGPT and allow you to write your next great app using Jumbo's chatbot!
This workshop is the antidote. Using TanStack AI - which works the same across every framework - you'll build a real, streaming AI assistant and learn when, where, and how to reach for AI so it genuinely makes your product better instead of just ticking a box. This isn't a tour of API syntax; it's the judgment that separates an AI feature people rely on from a demo that impresses once.
Everything you build transfers to whatever stack you ship on, and you'll walk away ready to start adding sensible, production-grade AI to your own apps the very next morning.
Who is this workshop for
- Anyone who needs to build with AI and wants to do it well, not just fast
- Developers curious about AI who want a real, end-to-end mental model instead of scattered tutorials
- Teams tired of bolted-on chatbots who want AI that actually improves the product
- Anyone who can call an API but isn't sure how to make an AI feature trustworthy, useful, and shippable - on any framework
What you'll learn
- Wire up TanStack AI end to end - stream a model's reply to the browser, consume it on the client, and shape the assistant's behavior with system prompts and agent configuration
- Give the model real tools (searching your data, taking actions in the app) while keeping a human in the loop, confirming before it acts, and designing tools that are safe, lazy, and predictable
- Return typed, schema-validated structured output your UI can trust, stream partial results so the screen fills in instantly, and carry structured replies across a multi-turn conversation
- The sensible version of each pattern - the decisions that make AI features reliable in production, not just flashy in a demo
- Patterns that aren't locked to one stack: because TanStack AI works across every framework, everything you learn here comes with you wherever you build
Required experience
- Comfortable with JavaScript and TypeScript
- Familiarity with any modern component framework - the examples run in a real app, but the patterns and TanStack AI itself are framework-agnostic
- No prior AI / LLM experience needed: you bring the product sense, we'll bring the AI
