How we estimatetime saved.
Your Edge page shows an estimate of the time your agent saved finding expert guidance. It is a deliberately low estimate from a public table, not a measurement of your work.
Table last updated 2026-09-29. These are editorial assumptions pending a timed study.
What the number means
For each eligible skill use, Edge adds the low-end minutes a person would spend finding, vetting and applying comparable expert guidance by hand, for that kind of task. It does not estimate the time to finish the whole task, and it does not measure what you would have done without Edge.
The category table
| Category | Assumed manual workflow | Low-end estimate | Last updated |
|---|---|---|---|
| Sales email | Find sales-email guidance, check fit to the audience, and apply it to a sequence. | 20 min | 2026-09-29 |
| Pitch deck | Find pitch-deck guidance, check the structure against the audience, and apply it to slides. | 30 min | 2026-09-29 |
| Data analysis | Find comparable analysis guidance, check the method against the data, and apply it. | 25 min | 2026-09-29 |
| Security review | Find security-review guidance, check applicability, and apply the checklist. | 30 min | 2026-09-29 |
| Frontend design | Find interface guidance, check the pattern against the product, and apply it. | 25 min | 2026-09-29 |
| Video production | Find production guidance, check the format and audience, and apply it to the piece. | 30 min | 2026-09-29 |
| Software development | Find implementation guidance, check it against the stack, and apply it. | 20 min | 2026-09-29 |
| Office document | Find document workflow guidance, check the file format, and apply it. | 15 min | 2026-09-29 |
Machine-readable copy: time-estimates.json. Edge's backend uses exactly these values.
What counts
- Eligible skill uses. Estimated time saved is based on skill uses that were loaded and not subsequently marked unsuccessful or abandoned. Most skill uses are not rated, so unrated eligible uses are included. A use your agent rated as no difference, not used, the wrong skill, broken, got in the way or unclear adds nothing. Helped and partly helped uses count.
- Abandoned uses and retries do not count. If your agent searched again for the same need and loaded a different skill instead, the earlier unrated skill adds nothing. Loading the same skill again for the same need counts once. Two skills used for different parts of one task both count.
- A narrow category match. The category comes from a narrow match on the search your agent made. A broad word like "data" or "design" is not enough. A use that matches no category adds zero minutes, so the total is a floor.
- Once per skill use. If your agent corrects a rating, the new rating replaces the old one. It never counts twice.
- Rounding. Under an hour the page shows minutes, such as ~35 min. From an hour it shows hours to one decimal, and whole hours from 10 h. It never shows 0.
- The same rule everywhere. "This week" covers the last 7 days. "Since you started" adds up every week since your history began, and carries the same (est.) marker.
The other numbers on Your Edge
Tasks reported improved
Tasks where your agent reported that a skill helped or partly helped, out of the tasks it rated. Your agent's own assessment, not an independent check.
Average impact
The mean of your agent's 0 to 10 ratings of the skill's effect. Shown only from 3 ratings.
Expert skills used
Distinct skills your agent loaded through Edge. Each one is security-checked and pinned to a reviewed version.
Streak
Consecutive days with at least one skill use. Shown from 2 days.
Changes against last week are shown as plain differences, such as +3 tasks or +0.6, and only when both weeks have the data for that number. Edge never shows a percentage change on a small base.
