# Blueprint: Thumbnail & title lab

URL: https://insidethemachine.ai/build/thumbnail-title-lab
Part: B-03. Family: Content studio. Status: Blueprint, not run yet.
Blueprint. Not run yet. First run on the Mac Studio from 10 Nov 2026.

Make many title and thumbnail ideas on your machine, score them all, and let Claude pick the few worth testing.

- Input: A finished video and a few lines about it.
- Output: 3 title options and 2 thumbnail ideas, each with a reason.
- Runs from: 8 GB of memory.
- Tools: ollama (https://insidethemachine.ai/setups/ollama.md), comfyui (https://insidethemachine.ai/setups/comfyui.md), draw-things (https://insidethemachine.ai/setups/draw-things.md)

## Claude's job

- Writes a short brief once: the angle, who it is for, and three title rules.
- Picks the top 3 titles and 2 thumbnail ideas from the scored list, and says why.

What Claude never sees: Claude sees the brief and the scored shortlist, not the video.

## Local parts per memory size

Data: models.json 2026-10-15, speech.json 2026-10-15, images.json 2026-10-15.

| memory GB | Drafter | Decider (Local Jev) | Image maker |
|---|---|---|---|
| 8 | Qwen3.5 4B (needs 5.8 GB, runs, but tight) | Qwen3.5 4B (needs 5.8 GB, runs, but tight) | Z-Image Turbo (needs 4 GB, runs well) |
| 16 | Gemma 4 12B (needs 11.2 GB, runs, but tight) | Qwen3.5 0.8B (needs 2.6 GB, runs well) | Z-Image Turbo (needs 4 GB, runs well) |
| 24 | Gemma 4 12B (needs 11.2 GB, runs well) | Qwen3.5 4B (needs 5.8 GB, runs well) | Z-Image Turbo (needs 4 GB, runs well) |
| 32 | Qwen3.8 27B (needs 22.7 GB, runs, but tight) | Qwen3.5 4B (needs 5.8 GB, runs well) | Ming-Image 0.1 Design (needs 24 GB, runs, but tight) |
| 48 | Qwen3.8 27B (needs 22.7 GB, runs well) | Gemma 4 12B (needs 11.2 GB, runs well) | Ming-Image 0.1 Design (needs 24 GB, runs well) |
| 64 | Qwen3.8 27B (needs 22.7 GB, runs well) | Gemma 4 12B (needs 11.2 GB, runs well) | Ming-Image 0.1 Design (needs 24 GB, runs well) |
| 96 | Qwen3.8 27B (needs 22.7 GB, runs well) | Qwen3.8 27B (needs 22.7 GB, runs well) | Ming-Image 0.1 Design (needs 24 GB, runs well) |
| 128 | Qwen3.8 27B (needs 22.7 GB, runs well) | Qwen3.8 27B (needs 22.7 GB, runs well) | Ming-Image 0.1 Design (needs 24 GB, runs well) |
| 192 | Qwen3.8-Flash-Next (125B-A6B) (needs 140 GB, runs, but tight) | Qwen3.8 27B (needs 22.7 GB, runs well) | Ming-Image 0.1 Design (needs 24 GB, runs well) |
| 256 | Qwen3.8-Flash-Next (125B-A6B) (needs 140 GB, runs well) | Qwen3.8 27B (needs 22.7 GB, runs well) | Ming-Image 0.1 Design (needs 24 GB, runs well) |
| 512 | Qwen3.8-Flash-Next (125B-A6B) (needs 140 GB, runs well) | Qwen3.8 27B (needs 22.7 GB, runs well) | Ming-Image 0.1 Design (needs 24 GB, runs well) |

How each part is picked:
- Drafter: Writes the first draft of many small things: summaries, title options, reply drafts. Rule: The smartest chat model that fits your usable memory and has an Ollama download (the same model the /do recipes install). (models.json)
- Decider (Local Jev): Answers typed questions about each item: pick one, give a score, or yes/no. It never writes. Rule: The smartest chat model with an Ollama download that needs no more than a third of your usable memory. Smaller models answer sooner, and a decider only picks a label. If no model is that small, the drafter's model decides too. (models.json)
- Image maker: Draws rough thumbnail ideas from a short prompt. Rule: The best-rated image model that fits and whose license allows commercial use. A YouTube thumbnail is commercial use. (images.json)
- Fit: We count 75% of your memory as usable for AI. The rest is for the system and your apps. Runs well: the model fits in usable memory with at least 10% to spare. Runs, but tight: it fits, but less than 10% of usable memory is left. Close other apps first. Needs N GB+: it does not fit. N is the smallest memory size it fits in. Size unknown: we do not know how much memory it needs.

## Steps

1. **Claude writes the brief** (Claude; stage id `brief`): The angle, the viewer, three rules for titles. A few hundred words, once per video.
2. **Draft 40 titles** (Your machine, Drafter; stage id `titles`): The drafter writes 40 title options that follow the brief.
3. **Score every title** (Your machine, Decider (Local Jev); stage id `score`): Local Jev scores each title for clarity and curiosity, and asks: does it promise something the video does not deliver (yes/no)? Those are dropped.
4. **Sketch thumbnail ideas** (Your machine, Image maker; stage id `images`): The image model draws rough ideas from 3 prompts. Ideas only: the final thumbnail is yours.
5. **Claude picks the shortlist** (Claude; stage id `choose`): 3 titles and 2 thumbnail ideas, each with a one-line reason.
6. **Make it yours** (You; stage id `yours`): Make the real thumbnail with your own face or footage. Never use a title the video does not deliver.
7. **Share your run (optional)** (You; stage id `share`): At the end, the script shows exactly what it would upload: your chip, memory, the models, how long each stage took and the token counts. No prompts, no outputs, no file names. Say yes and it goes to our review queue. Nothing is public until we approve it.

## Make it yours

- A title must match the video. Misleading titles and thumbnails break YouTube's rules.
- Use the AI sketches as ideas, then make the real thumbnail yourself.

## Receipts

- Claude tokens, this recipe: not run yet
- Claude tokens, all in Claude: not run yet
- Local minutes: not run yet
- Claude dollars (estimate): not run yet

No approved runs yet. Blueprints now, numbers later. We run every recipe on a 128 GB Mac Studio from 10 Nov 2026, then publish the receipts.

## Share your run

Opt-in. The script shows exactly what it would upload and asks first. It sends: Your chip, memory, graphics card and operating system; The recipe and the models you used; How long each stage took; Token counts: Claude in and out, local in and out; How many items came out; Tool versions; A fingerprint (SHA-256) of the log and up to 15 cleaned log lines; A handle you pick (not your name or email). Never: Your prompts or the model's answers; Your files, file names or folder paths; Emails, links, keys or IP addresses (the server rejects any upload that has one); Your IP address in plain form: we keep a scrambled copy for rate limits only. Nothing is public until a person approves it. Agents can submit with the `submit_run` tool of https://insidethemachine.ai/mcp.

Privacy: https://insidethemachine.ai/privacy.md