What I Use Instead

You are helping me turn one specific YouTube video into real, durable learning.

Source video:
- Title: What I Use Instead
- URL: https://www.youtube.com/watch?v=DogTO1jjFtI
- Topic: Creative Automation
- My current learning frame: Set up Pi with a local model, create an agents.md and one custom prompt file, then have Pi extend itself with a TypeScript safety-gate extension and verify it blocks a destructive rm command until you approve it.
- Why this matters: New playlist item from Web Dev Simplified; queued for transcript-backed review, topic mapping, and a practical learning artifact.

Transcript anchors from this exact video:
- 0:00 / Evidence 1: "Pi is an absolutely incredible agent harness and it's very similar to something like Claude code or open code, but the big difference is that Pi is incredibly minimal and it's super customizable. And this means that you..."
- 3:00 / Evidence 2: "local model instead of the model that's being paid for. So, now we essentially have the basic setup for our Pi terminal. We can hook up model, we can run different commands with that model, but what exactly..."
- 5:34 / Evidence 3: "to anyone else. Now, one other thing that we can do is we can really easily change what the thinking mode is for our model as well as swap between models with different keyboard shortcuts. For example, hitting..."
- 7:08 / Evidence 4: "it'll give you all the information you could want. And the nice thing about Pi is it allows you to actually swap between sessions and do various things with your sessions incredibly easily. For example, I can type..."
- 9:53 / Evidence 5: "compact that context to make sure that you have a smaller context overall. Now speaking of context, one really nice thing about pie is if you have a file called agent.md, so we can just come in here..."
- 15:17 / Evidence 6: "actually set up that theme. Okay, so just finished actually creating that theme. If we go ahead, we look, we have that prompts folder, and inside there we have a theme folder that hopefully is set up correctly..."
- 19:20 / Evidence 7: "someone else, so you can use their work to go off of. Or if you want to be able to create it yourself, you can create it yourself. Because since Pi is so minimal, they don't have MCP..."

Video-aware target:
- Prompt lane: Local model/runtime
- Mechanism to extract: Identify why the local setup works or fails for this specific agent task, not whether local models are generally good.
- Artifact to produce: A local model/runtime fit sheet with hardware constraints, model choice, endpoint setup, task benchmark, and fallback rule.
- Artifact must include: hardware; runtime; model/quantization; endpoint; agent integration; benchmark/fallback

Your task:
1. Use the transcript anchors above as the primary source packet. If you add outside context, label it clearly as outside context and keep it secondary.
2. Create a source-check table with columns: timestamp, claim, transcript support, what the demo proves, confidence, and what still needs verification.
3. Extract the actual teachable mechanism from the video: Identify why the local setup works or fails for this specific agent task, not whether local models are generally good. Do not invent claims that are not supported by the title, lesson frame, or transcript anchors.
4. Build a reusable learning artifact: A local model/runtime fit sheet with hardware constraints, model choice, endpoint setup, task benchmark, and fallback rule.
5. Include:
   - a plain-English definition of the core idea
   - a diagram or structured model using this sequence: Task -> Hardware -> Model/quantization -> Runtime endpoint -> Agent tool loop -> Benchmark task -> Fallback
   - answers to these source questions: What machine/runtime is shown? | What task exposes the model limit? | What setup change improves the loop?
   - 3 concrete examples that apply the video idea to real agentic work, such as Ollama or LM Studio coding endpoint; MLX Apple Silicon runner; DGX-backed Hermes session
   - 2 failure modes the video helps prevent, chosen from the transcript evidence and these likely risks: using a local model like ChatGPT; ignoring latency/context limits; no benchmark task
   - a checklist for the next real workflow, focused on: task fit, runtime setup, latency/context, tool loop, fallback
   - one practical exercise with a clear done signal: Choose one real coding task and specify the pass/fail benchmark for a local model.
6. Add a "learning transfer" section: what changes in my workflow tomorrow if I actually learned this?
7. Add a "source check" section that cites which transcript anchor supports each major takeaway.

Quality bar:
- Make this specific to "What I Use Instead", not a generic Creative Automation essay.
- Tie each harness element to a transcript anchor that names a tool, state boundary, permission, model behavior, or verification step.
- Prefer operational examples, failure modes, and reusable artifacts over broad definitions.
- Call out uncertainty instead of smoothing over weak evidence.
- Avoid these generic drifts: local-model ideology; hardware specs without workflow fit; benchmarks unrelated to the actual task.
- If evidence is weak or missing, stop and say what transcript segment or timestamp needs review instead of guessing.
- Finish with a concise artifact I could paste into my learning app.