# Before/After Time Log

Track the time each task takes before and after the AI workflow. Two weeks of honest numbers gives you proof of what you saved — proof you will use in Stage 4.

**Task:** [TASK NAME]   **Workflow:** [WORKFLOW NAME]   **Started:** [DATE]

## How to use this log

1. **Baseline first.** Before you switch to the workflow, time the task the old way at least twice. Use real runs, not estimates. Estimates flatter you.
2. **Then run the workflow.** Time every run for two weeks. Include input prep, prompting, verification, and rework.
3. **Count everything.** If the AI output failed and you redid part of it by hand, that time goes in the Rework column. Hiding it only fools you.
4. **After two weeks, do the math** in the summary section.

## Baseline (the old way)

| Run | Date | Minutes, start to delivered |
|---|---|---|
| 1 | [DATE] | [MIN] |
| 2 | [DATE] | [MIN] |
| 3 (optional) | [DATE] | [MIN] |

**Average baseline minutes (A):** [A]

## Workflow runs

| Run | Date | Prep | Prompts | Verify | Rework | Total |
|---|---|---|---|---|---|---|
| 1 | [DATE] | [MIN] | [MIN] | [MIN] | [MIN] | [MIN] |
| 2 | [DATE] | [MIN] | [MIN] | [MIN] | [MIN] | [MIN] |
| 3 | [DATE] | [MIN] | [MIN] | [MIN] | [MIN] | [MIN] |
| 4 | [DATE] | [MIN] | [MIN] | [MIN] | [MIN] | [MIN] |
| 5 | [DATE] | [MIN] | [MIN] | [MIN] | [MIN] | [MIN] |
| 6 | [DATE] | [MIN] | [MIN] | [MIN] | [MIN] | [MIN] |

Drop run 1 from the average if it included prompt debugging. Note it, but don't let setup cost distort steady-state numbers.

## The math

- Average workflow minutes (B): [B]
- Minutes saved per run: A − B = [A−B]
- Runs per month: [N]
- **Hours saved per month: ([A−B] × N) ÷ 60 = [HOURS]**

## Quality check

Time saved means nothing if quality dropped.

- Errors caught in verification this period: [COUNT]
- Errors that got past you: [COUNT — be honest]
- Output quality vs. the old way: [WORSE / SAME / BETTER, AND HOW YOU KNOW]

## Honest-numbers rules

- Verification time always counts. Skipping it inflates savings and ships errors with your name on them.
- Expect run 1 to be slower than baseline. That's normal.
- If savings are near zero after two weeks, refine the weakest prompt or return the task to manual. A dead workflow is a finding, not a failure.
- One good month is proof for you. Three consistent months is proof for an employer or client.

## Five-workflow summary

Fill this in when all five workflows have two weeks of runs.

| Workflow | Hours saved / month |
|---|---|
| 1. [NAME] | [HOURS] |
| 2. [NAME] | [HOURS] |
| 3. [NAME] | [HOURS] |
| 4. [NAME] | [HOURS] |
| 5. [NAME] | [HOURS] |
| **Total** | **[HOURS]** |

Part of the ADAPT framework — vetslearn.com/framework