How to analyze a 45-minute video with AI

A long recording is easier to review when you have a question, a timestamped first pass and a way back to the original footage. AdScanVideo Premium accepts uploaded videos up to 45 minutes. Here is how to turn that allowance into useful notes.

Start with one decision, not “analyze everything”

A 45-minute interview can contain dozens of topics. Decide what you need from it before uploading: a chapter outline, production notes, a shortlist of relevant moments or a summary of the speaker’s argument. A concise report can help you choose what to replay; it cannot prove that every small detail was captured.

Interview: Identify the main topics and the timestamps where each begins.
For each topic, summarize the speaker’s point and flag uncertain wording.

Product demo: Find moments where the dashboard is visible AND the speaker
explains an export. Return a short candidate list and explain the match.

Workshop: List the demonstrated steps in order. Separate visible actions
from spoken recommendations. Mark prerequisites that the video does not show.

For a chronological outline choose Summary; for an interview choose Podcast notes; for a focused search choose Find moments. If you need production references, Shot breakdown focuses on framing, camera movement and visible action. See our video summary guide for the difference between a synopsis and evidence-backed notes.

Prepare a 45-minute file under 200 MB

Duration and file size are separate limits. A 45-minute phone or camera recording can easily exceed 200 MB. Export a review copy from your editor rather than uploading a large production master. MP4 is a practical starting point; the uploader also accepts MOV, WebM and MKV.

For planning, 200 MB spread over 2,700 seconds is roughly 0.59 megabits per second for the entire file, before allowing for container overhead. Aim below that total, then check the actual exported file size. This is a size calculation, not a universal quality setting. Speech, screen text and fast motion need different compromises.

• Interview or podcast: keep speech clear; reduce unnecessary picture resolution before damaging audio.
• Screen recording: preserve readable labels. If the interface becomes too small, split the recording into focused parts instead of compressing it until text is unreadable.
• Fast action: preserve the original master and make a shorter review clip if compression hides the event you care about.

Remove private information you do not need to analyze, and upload only material you own or have permission to use. Retain the original file for verification.

The upload-to-report workflow

1. Sign in with email, then subscribe. Open pricing, choose Monthly Premium, verify your email through its sign-in link and complete Stripe checkout. Use the same email to restore your account in another browser.
2. Upload your review copy. Select a file up to 45 minutes and 200 MB. Platform links are not accepted.
3. Choose the report and focus. Select a mode and enter a specific question when the interface offers it. Avoid asking for an exhaustive transcript, every frame and a creative brief all in one pass.
4. Wait for the report. Upload, preparation and generation all contribute to processing time. A short test’s speed is not a promise for every long video.
5. Replay candidate moments. Use the original uploaded video and report timestamps to check what the AI observed. Correct uncertain quotes or timing before sharing.
6. Export your reviewed notes. Copy the report or export it to Word or PDF. Keep the source filename and a review date with your final document.

A worksheet for a long interview

This is an illustrative review worksheet, not an actual generated report. Imagine a 45-minute interview about an ad campaign. Use it to structure your human review after the first pass.

Source filename / duration:
Analysis mode / question:

Topic or candidate moment:
Report timestamp:
Verified timestamp in original:
What is visible:
What is audible:
What remains uncertain:
Decision or follow-up:

Reviewed by / date:

What to verify before using the report

AI video analysis samples and interprets audiovisual information. Brief actions, overlapping speech, small text and similar repeated events can be missed or confused. Timestamps are navigation aids; inspect the exact boundaries in the original. Do not describe this as exhaustive frame-by-frame analysis. Our accuracy and frame-by-frame review guide includes a more detailed verification worksheet.

For dialogue you intend to quote, replay it and verify every word. For Dashcam timeline, use descriptions to organize visible events, then review the original footage. A generated report does not determine fault, legal responsibility or measured speed. Read our methodology and limitations and data handling policy before uploading.

Frequently asked questions