Repurpose One Video or Webinar into Posts, Newsletter, Clips, and a Blog Outline
An agent turns one long video transcript into LinkedIn posts, a newsletter section, timestamped clip suggestions, and a blog outline in your brand voice, with a claims check and human approval before publishing.
The problem
Small businesses put real effort into webinars, podcast episodes, workshops, and long videos, then publish them once and move on. The ideas inside could fuel weeks of posts, a newsletter, and a blog article, but repurposing takes hours of rewatching, note-taking, and rewriting that nobody has time for. When it does happen, it is often rushed: a quote is taken out of context, a product claim slips in that the business cannot back up, or a guest's words get posted without checking they agreed. This recipe uses an agent to do the heavy first pass from the transcript, matching your brand voice guide, pointing to exact timestamps for clips, and flagging every factual or performance claim for review. A person approves each piece and confirms rights and consent before anything is published.
Who it's for
- Businesses that record at least one long video, webinar, podcast, or workshop per month
- Owners or small marketing teams who want consistent social and newsletter content without a full-time writer
- Teams with a written brand voice guide, or willing to create a one-page version
- Anyone who interviews guests or customers on recordings and needs a consent-aware process
Not a fit when
- Content in regulated areas such as health, financial, or legal advice where every claim needs professional compliance review
- Recordings where you do not own the footage or lack permission from participants to reuse it
- Businesses that want fully automatic posting with no review step
Tools
| Transcription tool (built-in platform captions or a dedicated service) | Produces a timestamped transcript from the recording |
| Claude or another LLM | Drafts posts, newsletter copy, clip suggestions, and blog outline from the transcript |
| Brand voice guide (Google Docs) | Defines tone, words to use and avoid, and formatting rules the agent follows |
| Google Sheets or a content calendar tool | Tracks each draft piece, its status, claims flags, approver, and publish date |
| Video editor (any desktop or browser editor) | A person cuts the suggested clips after reviewing timestamps |
| Google Drive | Stores the recording, transcript, consent records, and approved drafts |
| LinkedIn and your email newsletter platform | Publishing destinations, used only after approval |
Inputs you need first
- The recording and a timestamped transcript, ideally with speaker labels
- A one-page brand voice guide with tone, sample sentences, and banned phrases
- A list of approved product facts and claims you can support
- Guest or participant consent records covering reuse of their words and likeness
- Confirmation of rights to any music, slides, or third-party footage in the recording
- Target channels and how many pieces you want per recording
Step-by-step build
Write a one-page brand voice guide
In a Google Doc, describe your voice in three to five traits with an example sentence for each, list words and phrases to avoid, and set formatting rules: post length, emoji use, hashtags, call to action style. Paste in two or three past posts you liked as examples. This document goes to the agent with every run, so keep it short and specific rather than aspirational.
Create an approved claims list
List the factual statements about your business and products that you can back up: features, pricing ranges, guarantees, certifications. Note anything that must never be said, such as outcome promises or comparisons with named competitors. The agent will flag any claim in its drafts that is not on this list, so the reviewer knows exactly what to verify.
Confirm rights and consent before processing
Before a recording enters the workflow, check off in your tracker: you own or are licensed to reuse the footage, any background music is licensed for social and web use, slides or third-party clips shown are yours or permitted, and every guest has agreed in writing to reuse of their words and image in clips and posts. If consent is missing for a guest, mark them so the agent excludes their segments.
Produce a clean transcript
Generate a transcript with timestamps and speaker labels using your platform's captioning or a transcription service. Skim it and correct names, product terms, and obvious errors, because the agent will repeat mistakes it finds. Save it to the recording's Drive folder with a consistent filename so the automation can find it.
Run the repurposing prompt
Send the transcript, voice guide, claims list, and any excluded speakers to the agent with the prompt below. Ask for a fixed set of outputs: three to five LinkedIn posts, one newsletter section of 200 to 300 words, five to eight short clip suggestions each with start and end timestamps and a one-line hook, and a blog outline with headings and key points. Each piece must reference the timestamp range it came from.
Run a claims and quote check
Require the agent to append a review table: every factual claim, statistic, or promise in the drafts, whether it appears on the approved claims list, and the transcript timestamp where it was said. Also list every direct quote with speaker and timestamp. Anything not on the approved list is marked 'verify'. The reviewer works through this table first.
Log drafts in the content tracker
Write each piece to a Google Sheet with columns: recording, piece_type, channel, draft_text, source_timestamps, claims_flagged, speakers_quoted, consent_ok, approver, status, publish_date. Status starts as draft. This gives you one place to see what is pending and prevents an unapproved piece from being scheduled by mistake.
Human edit and approval
A named reviewer reads each piece against the source timestamps to make sure quotes are in context, resolves every 'verify' claim by confirming or removing it, and checks the speaker consent column. They edit for voice and mark approved. A person then cuts the suggested clips in a video editor, checking the actual footage for anything unsuitable such as visible screens, client names, or unlicensed audio.
Test with one past recording
Before using this on new content, run it on a recording you already repurposed by hand. Compare the agent's drafts with what you published. Check timestamps land on the right moments, quotes are accurate, and the claims table caught everything. Adjust the voice guide and prompt based on what felt off, then run once more.
Publish and maintain
Schedule approved pieces across the following weeks rather than all at once. After a month, review which pieces performed well on each channel and add those as examples to the voice guide. Update the approved claims list whenever products or pricing change, and remove pieces from the schedule if anything they say becomes outdated.
Example agent instructions
Adapt this to your business. Placeholders in braces are filled by your automation tool.
You repurpose long-form content for {business_name}. Follow the brand voice guide exactly: {voice_guide}. From the timestamped transcript below, produce: 1) {post_count} LinkedIn posts, each 80 to 200 words, built around one idea from the recording, with the source timestamp range noted. 2) One newsletter section of 200 to 300 words with a short headline. 3) Five to eight short clip suggestions, each with start and end timestamps, speaker, a one-line hook, and why it works standalone. 4) A blog outline with a working title, 4 to 6 headings, and key points under each, citing timestamps. Rules: Only use ideas actually said in the transcript. Do not use segments from these speakers, who have not consented to reuse: {excluded_speakers}. Quote speakers word for word or paraphrase clearly; never put words in quotation marks that were not said. Do not add statistics, results, or promises that are not in the transcript. At the end, add a review table listing every factual claim and every direct quote with speaker, timestamp, and whether the claim appears on this approved list: {approved_claims}. Mark anything not on the list as 'verify'. Transcript: {transcript_text}.
Human sign-off points
- Owner or marketing lead confirms footage rights, music licensing, and guest consent before the recording is processed
- A named reviewer approves every post, newsletter section, and blog outline before scheduling
- The reviewer resolves every 'verify' item in the claims table, confirming or removing it
- A person reviews actual footage for each clip before it is cut and published
- The guest is shown any clip or post featuring them before publishing when your consent terms require it
Guardrails
- Recordings without confirmed rights and consent do not enter the workflow
- Speakers without consent are excluded from drafts and clip suggestions
- Every draft piece references the source timestamp range so context can be checked
- Claims not on the approved list are flagged 'verify' and cannot be published unresolved
- Quotation marks are used only for words actually said, checked against the transcript
- Nothing is scheduled or posted until its status in the tracker is approved by a named person
- Clips are cut by a person who checks for unlicensed audio, private screens, or client information
Expected time saved
Estimate: if repurposing one hour-long recording by hand takes around 4 to 6 hours of rewatching, note-taking, and writing, an agent first draft with timestamps and a claims table might bring that down to roughly 1.5 to 2.5 hours of review, editing, and clip cutting. Results depend on transcript quality, how well your voice guide captures your style, and how many pieces you produce.
How to tell it's working
- Pieces published per recording, compared with your previous approach
- Reviewer editing time per piece
- Number of 'verify' claims per run, which should fall as the approved list matures
- Channel engagement on repurposed pieces relative to your usual posts
FAQ
Do I need permission to repurpose a guest's words from my own webinar?
Get explicit written consent that covers reuse in clips, posts, and articles before or at recording time. A simple release in your registration or booking form is a common approach. Consent rules vary by location and context, so if you are unsure, ask a professional. This recipe excludes speakers without recorded consent by design.
Can the agent pick the best clips automatically?
It can suggest moments with timestamps and hooks based on the transcript, but it cannot see the video. A strong line might be paired with poor audio, someone looking away, or a screen showing private information. A person should watch each suggested segment before cutting it, which is usually quick once the timestamps are provided.
How do I stop the posts sounding generic?
Give the agent a specific voice guide with real examples of your past posts, banned phrases, and formatting rules. Ask for posts built around one concrete idea from the recording rather than broad summaries. Edit drafts with a light hand at first and add your best final versions back into the guide as examples.
Want this set up for you?
Build it yourself with the steps above, or tell us what you need. We scope the work with you and send a written quote before anything starts.
Sources
Written by Agents Lookup. Official pages for the tools this recipe uses:
- Claude (official page): claude.ai (checked 2026-11-09)
- Google Drive (official page): developers.google.com/workspace/drive (checked 2026-11-19)
- LinkedIn Hiring Pro (official page): business.linkedin.com/hire/hiring-pro (checked 2026-11-19)