Short Technical Videos Worth Your Time

Not all videos are created equal. We maintain a short, opinionated list of videos to watch when you want to learn fast.

Start here: build real copilots and agents

• Mastra AI — official channel

  • Channel: https://www.youtube.com/@mastra-ai
  • Why watch:
    • TypeScript‑first framework for agentic apps and copilots.
    • Clear examples of tools, orchestration, eval hooks, and RAG integrations.

• CopilotKit — official channel

  • Channel: https://www.youtube.com/@CopilotKit
  • Why watch:
    • In‑app copilots with UI bindings, function calling, and server actions.
    • Patterns for productizing assistants inside web apps.

• E2B — official channel

  • Channel: https://www.youtube.com/@e2b-dev
  • Why watch:
    • Secure sandboxes and managed runtimes for code‑executing agents.
    • Useful when your workflow needs safe, isolated tool execution.

Pairs well with: our coding agent reference architecture → /blog/build-a-coding-agent-ts

How we choose

• Durable insights over hype; production over demos.
• Clear mental models and concrete takeaways you can apply this week.
• Minimal vendor lock-in; when a tool is shown, we explain the general pattern.

Use this list like a syllabus: pick one per section, take notes in a doc, then ship one change influenced by what you learned.

More to watch

Agents and tool use

• State of GPT — Andrej Karpathy

  • Watch: https://www.youtube.com/watch?v=xO73EUwSegU
  • Why watch:
    • A big-picture map of how GPT-style models work and where they’re going.
    • Frames how reasoning, tools, and training advances fit together.
  • Pairs well with: our post on agent loops that actually ship value → /blog/beyond-autocomplete

• LangChain Agents in 2025 (v0.3)

  • Watch: https://www.youtube.com/watch?v=Gi7nqB37WEY
  • Why watch:
    • Modern agent APIs and how tools, memory, and control flow fit together.
    • Practical patterns for guarded tool use and observability hooks.

RAG: from basics to production

• Learn RAG From Scratch — a step-by-step tutorial

  • Watch: https://www.youtube.com/watch?v=sVcwVQRHIc8
  • Why watch:
    • Walks through retrieval pipelines end to end with code.
    • Great foundation if you’re building your first RAG workflow.

• Practical Tips for Building Production-Grade RAG (Jerry Liu, LlamaIndex)

  • Watch: https://www.youtube.com/watch?v=TIouyATCHbU
  • Why watch:
    • Concrete guidance on chunking, indexing, and query-time composition.
    • Focus on failure modes you’ll actually hit in prod.
  • Pairs well with: our open-source RAG guide → /blog/rag-for-your-blog-open-source

Vector databases and embeddings

• What’s in a Vector Database? (JP Hwang, Weaviate)

  • Watch: https://www.youtube.com/watch?v=flVddNTB-Gs
  • Why watch:
    • Clear mental model for vectors, similarity, and index structures.
    • Helps you reason about recall/latency tradeoffs.

• Pinecone Vector Database Explained

  • Watch: https://www.youtube.com/watch?v=OvjGhsCyakQ
  • Why watch:
    • Solid intro to embeddings, search, and practical usage patterns.
  • Pairs well with: our post on what to measure in production → /blog/vector-databases-in-production

Observability, tracing, and evals

• LangSmith 101: Observability for AI Apps

  • Watch: https://www.youtube.com/watch?v=Iyc80hY2yYk
  • Why watch:
    • Tracing prompts, tool calls, and outputs; building feedback loops.
    • A reference for what “good” instrumentation feels like.

• Phoenix (Arize) for LLM/RAG/Agent tracing and evaluation

  • Watch: https://www.youtube.com/watch?v=5PXRRXM8Iqo
  • Why watch:
    • Open-source workflow for experimenting, evaluating, and troubleshooting.
    • Useful even if you don’t adopt Phoenix—copy the patterns.

• RAG Evaluation with RAGAS

  • Watch: https://www.youtube.com/watch?v=mEv-2Xnb_Wk
  • Why watch:
    • Practical metrics and methods to move from vibes to evidence.
    • Shows how to check context precision/recall and answer faithfulness.

Fundamentals that pay dividends

• Let’s build GPT from scratch, in code (Karpathy)

  • Watch: https://www.youtube.com/watch?v=kCc8FmEb1nY
  • Why watch:
    • Ground-up tour of attention and training that clarifies the black box.
    • Not required for app work, but superpowers your debugging and intuition.

• DSPy: Transforming LM Calls into Smart Pipelines (Omar Khattab)

  • Watch: https://www.youtube.com/watch?v=NoaDWKHdkHg
  • Why watch:
    • A principled approach to prompting, program synthesis, and evals.
    • Think of this as “software engineering for prompts and pipelines.”

Industry context

• OpenAI DevDay: Opening Keynote
  • Watch: https://www.youtube.com/watch?v=U9mJuUkhUzk
  • Why watch:
    • Fast overview of platform capabilities, pricing shifts, and new primitives.
    • Helps you decide when to build on vendor services vs open-source.

Bonus channels

• How I Built This with AI — channel

  • Channel: https://www.youtube.com/@howiaipodcast
  • Why watch:
    • Founder and engineer interviews focused on shipping real AI products.
    • Practical stories about agents, RAG, evals, and go-to-market.

• LangChain — official channel

  • Channel: https://www.youtube.com/@LangChain
  • Why watch:
    • Tutorials on LangChain, LangGraph, and LangSmith.
    • Good for staying current with API changes and best practices.

How to use this list

• Pick one video from each section across a month.
• After each watch, add a 5–10 line note to your team docs: what changed in your understanding and what you’ll try.
• Ship one small improvement per week influenced by a video (a check, a log, a prompt fix, or a guardrail).

See also

• Beyond autocomplete → agent loops that do real work: /blog/beyond-autocomplete
• Build a RAG search for your blog: /blog/rag-for-your-blog-open-source
• Vector databases in production: what to measure: /blog/vector-databases-in-production
• Reference coding agent architecture: /blog/build-a-coding-agent-ts