From the Trenches is a Blastra series featuring practitioners solving real problems at work - one shot at a time.
About a year and a half ago, Files.com held a two-day customer event. Speakers presented, sessions were recorded, and by the end of it we had hours of video with real customer stories and use cases.
I should mention: at the time, I was the entire marketing team on this project. Files.com is a file orchestration platform used by more than 4,000 organizations, and we had plenty of product and engineering muscle, but marketing was brand new. At most companies, these recordings would end up on an old file server collecting digital dust. At Files.com, they landed on my desk with no team to hand them to.
I looked at what was probably 18 hours of video and thought that I would build a workflow using AI.
Four Steps, Three Tools
Step 1: Video to text. I used ChatGPT directly to ingest the video recordings along with the presentations the presenters used - feeding in the audio and getting back usable transcripts. The videos themselves weren't stellar content. They were a single tripod setup in the back of the room. The audio was clear, but it wasn't engaging or compelling, and I knew there was good material in the content being delivered.
Step 2: Transcript to draft. With the transcripts in hand, I used ChatGPT to reshape the raw presentations into structured blog post drafts. A 45-minute session doesn't translate directly into a readable post - there's back-and-forth, tangents, context that only makes sense live. The AI handled the first pass of pulling out the key points and organizing them into something publishable. The blog drafts weren't publish-ready on first generation, but they were about 75% of the way there, and even that was a massive time-saver.
Video-to-blog isn't the most mind-breaking workflow though - I didn't stop there. I used the transcript and blog post to generate a concise four-to-five minute voiceover script that took the content from long-form lecture style to a compelling slide deck walkthrough, and prompted ChatGPT to create a matching slide deck outline I could import into Canva for refining (I didn't trust or expect ChatGPT to create something beautiful and on-brand).
Step 3: Voiceover. I used ElevenLabs to generate the voiceover from the script. I picked a voice similar to one that had done some knowledge base content for us previously because it felt like a good fit. I had to tweak some pronunciation, but that was expected with so many acronyms and technical jargon to handle.
Step 4: Visuals. With Canva I handled the visual side. I uploaded the ChatGPT-created deck as a starting point, then created my own template that I could apply to slides to make them more visually appealing and on-brand. I uploaded the ElevenLabs voiceover into Canva, set timing on each slide to match, and exported everything as a single .mp4 file.
Then - cracking my knuckles - make a YouTube thumbnail and publish.
What Came Out the Other End
Six months of content from two days of recordings. I standardized this workflow and made some custom prompts and GPTs to make the generation process even faster.
Here's an example of what came out of the pipeline:
You Don't Need a Team for This
I wasn't setting out to build a "content repurposing system" or an "AI content engine." I was one marketer with a pile of recordings and no help. The AI tools made it possible for me to do the work a content team would have done, at a pace that made sense. Now the process is similar, but I've discovered purpose-built tools that handle much of what I had Frankensteined together.
If you're sitting on event recordings, webinar archives, or customer call libraries and thinking "we should do something with those" - you probably should, and you probably don't need a lot of help to do it.
Have a story from the trenches? Drop us a line at ceo@blastra.io
Related Reading
- Who Decides What Your Software Is? — How software categories shape your visibility before buyers ever find you
- Introducing Visibility Posture: What Your Directory Presence Actually Looks Like — A new way to see where you stand across directories
- Who Owns Your Software Reviews? What the Fine Print Actually Says — What happens to your reviews when platforms change the rules