01How to read this
This is the working record of the choices made while producing your deliverable: what was explored, what was set aside, why, and where a person worked on the output by hand. Countersigning it at each milestone is what makes it a record both parties hold.
02lede
6 directions were generated and compared. 1 was set aside. Recorded reasons: Reads as an insurance ad, which is what the brief said to avoid. The direction taken: Opens on the customer's morning, not on the product. A person then worked on the result directly for 1.4 h.
human 73% · generated 27% · 6 decisions recorded · 1.4 h hands-on
| When | Decision | By | Recorded reason | Capture |
|---|---|---|---|---|
| 2026-08-24 09:45 | CONSTRAIN | OH | Client legal: no guarantee wording, no comparison to named insurers. | manual |
| 2026-08-24 10:10 | SPEC | OH | Six opening lines, Thai first, under 20 words each. | auto |
| 2026-08-24 10:11 | GENERATE | claude opus-5 | — | auto |
| 2026-08-24 10:30 | REJECT | OH | Reads as an insurance ad, which is what the brief said to avoid. | guided |
| 2026-08-24 10:34 | SELECT | OH | Opens on the customer's morning, not on the product. | guided |
| 2026-08-24 14:00 | INTERVENE | Ploy | Rewrote both sentences; kept the image, replaced every word. | guided |
03body_promise
4 directions were generated and compared. The direction taken: Second draft carries the promise without a claim legal would strike. A person then worked on the result directly for 2.2 h.
human 71% · generated 29% · 4 decisions recorded · 2.2 h hands-on
| When | Decision | By | Recorded reason | Capture |
|---|---|---|---|---|
| 2026-08-25 09:20 | SPEC | OH | Three paragraphs on the promise, plain Thai, no jargon. | auto |
| 2026-08-25 09:21 | GENERATE | claude opus-5 | — | auto |
| 2026-08-25 10:05 | SELECT | Ploy | Second draft carries the promise without a claim legal would strike. | guided |
| 2026-08-25 11:40 | INTERVENE | Ploy | Restructured into two paragraphs and rewrote the middle one entirely. | guided |
04body_proof
A person then worked on the result directly for 3.5 h.
human 100% · generated 0% · 2 decisions recorded · 3.5 h hands-on
| When | Decision | By | Recorded reason | Capture |
|---|---|---|---|---|
| 2026-08-25 15:00 | DIRECT | OH | Ploy to draft the proof section from the claims data, not from the model. | manual |
| 2026-08-26 13:15 | INTERVENE | Ploy | Written from source, no generated draft used. | manual |
05close
3 directions were generated and compared. The direction taken: Returns to the opening image without repeating the line.
human 100% · generated 0% · 2 decisions recorded
| When | Decision | By | Recorded reason | Capture |
|---|---|---|---|---|
| 2026-08-26 16:30 | SPEC | OH | Close on the same morning image as the lede. | manual |
| 2026-08-27 09:10 | SELECT | OH | Returns to the opening image without repeating the line. | guided |
06Milestones and countersignatures
| Milestone | Date | Countersigned by | Signed at |
|---|---|---|---|
| Brief accepted | 2026-08-24 09:30 | — | — |
| Concept approved | 2026-08-25 16:00 | act_client2 | 2026-08-25 17:05 |
| Final delivery | 2026-08-27 15:00 | — | — |