The ask

Show the client's leadership what AHEAD built.

The client, a global insurer, is one of AHEAD's flagship AI clients. Our team had built a tool for them, and the managing principal on the account wanted a demo he could put in front of the client's executives. The goal was simple: make the value obvious enough that the client would want to do more with AHEAD.

What I was handed

Screenshots, half a script, and a lot of good intentions.

The first call was "build a demo." What existed was a pile: screenshots, a few screen recordings, half a script, half a storyboard, and graphics that didn't match. Everything was true, but none of it told a story.

The decisions

Start over from one clean source.

1. Record the engineer, not the scraps. I sat with the engineer who built the tool and recorded his screen in Descript while he walked me through the tool, so I had his explanation and a transcript in one take.
2. Let Claude draft, then edit for the room. I gave Claude the transcript and screenshots. It built a storyboard and a new script and flagged what to focus on in each section. I rewrote it for an executive audience.
3. Rebuild the interface instead of showing raw recordings. I recreated the platform's screens and animated them, so every moment was clean, readable, and on message.

The build

A workflow I can run for any software demo.

1. 1Walkthrough recording
2. 2Transcript and screenshots
3. 3Claude storyboard and script
4. 4Animated rebuild and edit

Recording and transcribing in one session gave Claude everything it needed to draft. That left my time for the parts that need judgment: what an executive cares about, what to cut, and how each screen should move.

The result

Client-ready in three days.

The demo was shown to the client's leadership, and the feedback that came back was simple: the client thought the video was great. It took three days to build and one more for edits, with only a few rounds of revisions.

The problem

Make it move, and make it consistent.

I'm AHEAD's only video editor, and I was never formally trained in animation. Designing something once was fine. Making it move the same way for 30 to 120 seconds without anything breaking or looking silly was close to impossible. I'd try an idea, it wouldn't work, and I'd delete two to four hours of work. A single piece could take weeks.

The decisions

Teach the brand first. Let Claude build the bones.

1. Start from the brand. I built a design system in Claude from AHEAD's brand guidelines and taught it the motion styles I like. Every output starts on brand instead of needing a fix later.
2. Storyboard before anything moves. Claude draws the video scene by scene first. I critique the frames while changes are still cheap.
3. Pick the engine for the job. Short pieces Claude builds directly. Long or complex ones go through HeyGen HyperFrames.
4. Always land in After Effects. The final output is a script I run in After Effects, so I can adjust any keyframe, speed or timing. Claude builds the bones, and I perfect it.

The build

Five steps from idea to editable motion.

1. 1Idea and key points
2. 2Claude storyboard, scene by scene
3. 3My revisions
4. 4Claude or HyperFrames build
5. 5After Effects script, then my polish

In practice: the Ready, Set, Runtime ad. All we had was a rough recording of the show's host reading some ad copy, and we wanted a full 15-second ad for the podcast. I cut his read down to 15 seconds, drew a storyboard in Illustrator, and refined it with Claude. We picked the strongest frames to animate. Claude synced the motion to the voiceover and wrote the After Effects script, and I used that file to polish every move by hand.

The result

Weeks of work, now an afternoon.

The pipeline now powers the cold opens, graphics and ad for AHEAD's podcast, plus reusable title-card templates. For a new interview question card, I swap the text and render. The time I used to spend fighting keyframes now goes into the story.

The ask

Great footage. No defined deliverable.

AHEAD opened Foundry, a new facility built for AI workloads. On short notice, a colleague from marketing interviewed four of our experts on site, and I helped edit the questions beforehand. I spent a day on site shooting B-roll of the facility. Afterward I had hours of raw, unscripted answers. Nobody had decided what they should become.

The decisions

Find the story first. Then propose it.

1. Transcribe everything. I ran all four interviews through Descript so I could read the footage instead of scrubbing it.
2. Pull the strongest moments. I picked my favorite quotes and sent them over for a direction.
3. Keep it moving. Everyone was busy, and the direction didn't come back for two weeks. Rather than let good footage sit, I used Glean on the transcripts to shape each expert's quotes into its own story arc.
4. Ship one, then scale. I cut the first video with B-roll so the team could react to something real instead of an outline.

The build

From raw answers to a story people can follow.

1. 1Shape the interview questions
2. 2Descript transcripts of all four
3. 3Glean finds a story arc per expert
4. 4Cut video 1 with B-roll, publish with transcript

The result

The transcript idea worked.

The video and its transcript went onto AHEAD's Foundry page. In September, our search agency flagged that page as one of AHEAD's top AI citations in Google search. Marketing passed the note along:

"So the transcript idea seems to be working!"

I didn't build the page. My part was the content on it: the story, the cut and the transcript that search engines and AI tools could read.

The problem

Consistent thought leadership, from one editor.

AHEAD's experts have a lot to say, but there was no steady place for them to say it. An earlier podcast attempt had stalled. When marketing brought the idea back, the show's leads set the format. Making it happen every episode, with one editor, was my job.

The decisions

Design the system before episode one.

1. Plan for everything a recording can become. The full episode is only the start. Clips, Shorts, a blog and guest assets all come from the same hour.
2. Let the transcript drive. Once I've checked the Descript transcript, it feeds the descriptions, chapters, cold-open ideas and the companion blog.
3. Use Claude for options, and keep the judgment mine. Claude pitches three cold-open storylines. I pick one and cut it.
4. Give the guests something. Every guest gets their own clips to share, which grows the audience for free.

The build

One recording, ten-plus outputs.

The result

A show that runs on a system, not on heroics.

Three episodes in, Ready, Set, Runtime is among AHEAD's top-performing YouTube content. Each one ships with the full set of outputs, and the social team and guests get material they can use the same week. Adding an episode adds work, but it doesn't add chaos.

• Episode 1What is Accelerated Dev, and Why Does it Change Everything?Watch on YouTube ↗
• Episode 2The Gap Between the Demo and the DeployWatch on YouTube ↗
• Episode 3The New SDLCWatch on YouTube ↗