How to Make a Founder Story Video with AI
The 4-step workflow for AI founder-story ads: real-footage-plus-AI-b-roll and full-avatar paths, the 'why I built this' script arc, authenticity cues, and cost math vs a videographer day rate.
Founder-story ads work for DTC because they answer the question every skeptical buyer has: who is behind this, and why should I trust them. This guide gets you to a 30-to-45-second founder-led clip two ways: real founder footage stitched with AI b-roll, or a full AI avatar of the founder built with their consent. You'll get the "why I built this" script arc, the authenticity cues that keep it from feeling like a corporate spot, and the cost math against a videographer day rate. A one-day founder shoot with a videographer runs $800 to $2,500 plus scheduling around a busy founder. The AI-b-roll path can cost as little as $10 to $30 in credits on top of a phone recording.
TL;DR
- Step 1: Write the "why I built this" arc, problem the founder lived, the fix they made, the product it became.
- Step 2 (Path A): Record the founder talking on a phone, then generate AI b-roll (product, process, origin scenes) with Seedance 2.0 and Kling 3.0 to cover the story.
- Step 2 (Path B): With explicit founder consent, build a full avatar with Higgsfield Soul 2.0 locked to one reference portrait, and generate the talking head.
- Step 3: Generate origin and product b-roll to visually prove the story beats.
- Step 4: Assemble in a plain, un-polished cut, add captions, keep one honest imperfection.
Path A costs $10 to $30 on top of a free phone recording. Path B costs $8 to $20 fully synthetic. Both ship same-day.
Why founder-led beats a spokesperson
A founder is the one person who cannot be swapped out. A hired actor reads a script; a founder is the story. For DTC brands that live or die on trust, the founder's face and the specific, slightly awkward reason they started the company converts better than a polished testimonial because it cannot be faked by a competitor. That is also the reason the authenticity cues below matter more here than in any other format: over-polish is what kills a founder video.
The two paths trade off differently. Path A keeps the founder's real voice and face and uses AI only to cover the parts you cannot easily film (the early prototype, the factory, the 2am kitchen where it started). Path B replaces the shoot entirely with a consented avatar, useful when the founder cannot or will not sit for footage, or when you want to test the format before asking for their time.
The 4-step workflow
Step 1: Write the "why I built this" arc
No model yet. The arc has three beats and one rule.
- The problem you lived. Not a market problem, a personal one. "I have eczema and every 'gentle' cleanser I bought still burned."
- The turn. What you did about it. "So I spent a year working with a chemist to strip out the stuff that was actually causing it."
- The product. What it became, stated plainly. "That's [brand]. It's the cleanser I wished existed."
The rule: specificity over polish. "I was frustrated with the industry" is filler. "I returned $340 of skincare in one month" is a founder story. For this guide our example is a fragrance-free cleanser founded by someone with sensitive skin.
Step 2, Path A: Real footage + AI b-roll
Models: Seedance 2.0, Kling 3.0 for b-roll
Have the founder record the three beats on a phone, front camera, near a window, one continuous take per beat. This is the cheapest and most authentic anchor you can get. Then generate b-roll to cover the story visually so you can cut away from a static talking head.
Origin b-roll with Kling 3.0 (face-light, high volume):
Close-up hands mixing a cream in a small kitchen bowl at night, warm lamp
light, shallow depth of field, unpolished handheld feel, vertical 9:16,
4 seconds.
Product b-roll with Seedance 2.0 (upload the real product):
[Product reference] fragrance-free cleanser bottle on a bathroom shelf,
soft morning light, a hand reaches in and picks it up, vertical 9:16,
product stays in focus center frame, 4 seconds, authentic handheld framing.
The founder's real face carries the story; the AI b-roll carries the proof. Nobody has to know which shots are which.
Step 2, Path B: Full avatar with consent
Model: Higgsfield Soul 2.0
Only take this path with the founder's explicit, documented consent to create a synthetic likeness. This is their identity; treat it that way. With consent, upload one front-facing reference portrait and generate the talking head:
[Founder description] sitting at a kitchen table in soft natural window
light, looks directly at camera, says "[beat line]" with a calm, sincere
expression, occasional natural pauses. Vertical 9:16, handheld feel, no
music, clean audio.
Generate each of the three beats as its own clip, all anchored to the same reference so the founder's face stays consistent across cuts. For more on synthetic likenesses and how identity locking works, see what is an AI avatar.
Step 3: Generate origin and product b-roll
Models: Seedance 2.0 (product), Nano Banana Pro (stills), Kling 3.0 (process)
Whichever path you took for the talking head, you need cutaways that prove the story. For our cleanser: the early prototype, the ingredient close-ups, the finished product on a shelf. Generate product motion with Seedance 2.0 (real product reference), process shots with Kling 3.0 for speed, and any end-card or freeze-frame stills with Nano Banana Pro:
Fragrance-free cleanser bottle on a clean white background, soft studio
light, hero product shot, no text, premium and minimal.
Nano Banana Pro renders these in roughly 8 to 15 seconds. Use them as the end card behind the CTA.
Step 4: Assemble with one honest imperfection
Tools: 8frame Studio or any NLE
Cut it plain. A founder video that looks like it went through a color-grade suite reads as an ad; a founder video with one slightly uneven cut reads as real. Keep the founder's natural pauses. Add captions in white with a thin outline. A 40-second structure:
| Seconds | Shot | Content |
|---|---|---|
| 0 to 4 | Founder, talking head | The personal problem (this is your hook) |
| 4 to 12 | Founder + origin b-roll | The turn: what they did about it |
| 12 to 25 | Product b-roll + founder | What it became |
| 25 to 35 | Founder | Why it matters to them |
| 35 to 40 | Product still + CTA | "Made this for people like me. Link in bio." |
If you used Path B, the AI-generated label goes on at upload, same as any synthetic creative. A consented founder avatar is still AI, and the disclosure rules apply; the disclosure guide covers the specifics.
Cost math
Videographer founder shoot:
- Videographer day rate: $800 to $2,500
- Founder's time: half to a full day, the real cost for a busy founder
- Edit: $200 to $600
- Turnaround: 1 to 3 weeks including scheduling
- Output: one shoot, one story angle
Path A (real footage + AI b-roll):
- Phone recording: free, 20 minutes of the founder's time
- Seedance + Kling + Nano Banana b-roll: $10 to $30
- Turnaround: same day
- Output: multiple story-angle cuts from one recording session
Path B (full avatar):
- Higgsfield talking head across beats: $6 to $14
- B-roll: $4 to $10
- Turnaround: same day
- Output: test the format without spending founder time at all
Path A is the sweet spot for most DTC brands: real founder, real voice, AI only where filming is expensive. Path B is the testing lane. For the strategic tradeoff, see the AI UGC vs real creators playbook.
Common pitfalls
Over-polishing. The number-one killer. A graded, tripod-steady, studio-lit founder video loses the exact trust the format exists to build. Keep it handheld and slightly imperfect.
Generic origin story. "I couldn't find a product I loved" is every brand. The specific returned-$340 detail is what makes it a founder story. Write the specifics.
Using Path B without consent. Building a synthetic likeness of a real person requires their explicit, documented permission. This is non-negotiable, ethically and legally.
No captions. Sound-off viewing is the majority. If the founder's opening line is not on screen, most viewers never hear the hook.
FAQ
Should I use real founder footage or a full AI avatar?
Use real footage plus AI b-roll (Path A) whenever the founder can spare 20 minutes to record on a phone. It keeps their real voice and face, which is the whole value of the format, and uses AI only for the expensive-to-film b-roll. Reserve the full avatar (Path B) for testing the format quickly or for founders who genuinely cannot sit for footage, and only ever with their explicit consent.
Do I need consent to make an AI avatar of my founder?
Yes, always. Creating a synthetic likeness of a real person requires their explicit, documented consent, even when that person is your own founder. Beyond the ethics, the AI-generated content must carry the platform's AI label at upload. Treat the founder's likeness as the sensitive personal data it is.
Why do founder videos convert for DTC brands?
Founder-led videos answer the trust question that polished ads cannot: who is behind this and why. A founder telling the specific, personal reason they built a product is something a competitor cannot copy, and the slight awkwardness of a real founder reads as more credible than a smooth spokesperson. For DTC brands where trust drives conversion, that authenticity is the performance lever.
Write your three story beats, then record the founder on a phone for Path A or, with consent, generate the talking head for Path B. Clone a UGC template on 8frame's workflow library, or start on the canvas with every model above in one place.