What could the next shift make?
Potential hands-on time returned per week.
Clips × (manual time − review time) ÷ 3,600. Inputs are examples, not measured product results. Processing, setup, and transfer time are excluded.A useful first shift
A new student editor should be able to find a clean reaction, assemble a short sequence, and learn why a shot works. When the whole shift goes into opening files and typing the same labels, the department loses an opportunity to teach the work that made the student join in the first place.
A reviewed log can give that time back. Start with a narrow assignment, such as preparing one quarter of footage for a social edit, and define the questions the next editor needs to answer.
Set one vocabulary for the team
| Field | Type | Example use |
|---|---|---|
| Subjects | Known subject records | Athletes, staff, or other recurring people |
| Action | Single-select or Multi-select | The activity that matters to the story |
| Shot size | Single-select | Wide, medium, close-up |
| Attributes | Multi-select | Crowd, celebration, reaction |
| Pick | Flag | Worth returning to for this edit |
| Rating | Rating | Editorial usefulness under your team's rubric |
| Note | Text | Context the controlled fields cannot express |
These are example fields, not claims about a particular team's production. Keep the first palette short enough that the assistant will use it consistently. Expand it only when a real retrieval request exposes something missing.
Let suggestions handle the repeated description
Choose the reasoning route before running the batch. Local can suit a controlled or offline setup once its model is ready. Rush can suit a deadline where hosted execution is worth the cost. Your API uses your own configured provider with the required entitlement.
Review a varied sample first. Correct the wrong action, reject an uncertain person match, and add the context that the image cannot establish. Helmets and duplicate jersey numbers are reasons to inspect evidence, not reasons to assume the model knows the roster perfectly.
Measure editing time recovered
The worksheet at the top of this guide is an estimate you control. Enter your observed manual logging time and the proposed review time. It calculates hours potentially returned to the team; it does not claim that ClipLogger has achieved those savings or convert unpaid student time into a fictional wage saving.
Measure the next shift using the same folder and retrieval tasks. Keep model runtime separate from hands-on review and verify that another editor can use the resulting log. A faster first pass is only useful if the handoff remains accurate.
Finish with an actual edit
Have the assistant filter the reviewed log, create a small set of selects, and send them into the NLE. Ask the receiving editor which fields were useful and which were ignored. Use that feedback to refine the palette rather than adding more categories by default.
Start with the first session, roster setup, and reviewing suggestions.