LLenny's Podcast
← All frameworks
InnovationGaurav Misra (CEO and co-founder of Captions)

Documentation vs Storytelling Safety Framework

Sort every AI-video use case into documenting reality (block) or storytelling (enable)

Difficulty
Moderate
Time to result
~ongoing to results
Steps
3
Confidence
88%

A design principle for building generative-AI-video products responsibly. Divide all video into two categories. 'Documentation' captures what actually happened — personal memories, reporting, history — where a fake generated version has zero upside and only harm. 'Storytelling' is entertainment — ads, social posts, TV, movies — where everyone knows it's fabricated and enabling more people to tell stories is purely positive. Then deliberately design products to be hard to use for documentation and easy for storytelling.

Origin

Gaurav Misra's product-safety framework at Captions for generative talking-video.

Core principles

  • 01Generating fake versions of reality (documentation) has no legitimate benefit — only harm
  • 02Storytelling video is understood by its audience to be fabricated, so generating it is a pure positive (a Geico gecko, reality TV)
  • 03Product and design choices — not just policy — should structurally steer usage toward storytelling and away from documentation

How to run it

  1. 1

    Classify the use case

    Ask whether a given video is documenting real events (personal memory, reporting, history) or telling a story (ad, social post, TV, movie).

  2. 2

    Block or discourage documentation uses

    For anything that falls in the documentation bucket, design the product to make that use case difficult.

    Watch out Fabricating reality to fool people has no upside — treat it as strictly negative regardless of how exciting the capability is.

  3. 3

    Enable storytelling uses

    For storytelling, make the product easy and delightful to use, since empowering more people to entertain and share is a net positive.

In the wild

Focusing Captions on talking video for stories

Captions deliberately targets dialogue/monologue 'talking video' for storytelling contexts (ads, social, entertainment) and designs its products to be easy there while making reality-fabrication use cases hard.

A product-safety posture baked into design rather than bolted on as after-the-fact moderation.

Common mistakes

Treating safety as only a policy layer

Misra argues the real challenge — and the real lever — is in product and design choices that make misuse structurally hard, not just terms-of-service rules.

Is it for you?

Best for

Founders and product/design teams building generative-media tools who need a principled, buildable stance on misuse

Not ideal for

Products where the same output legitimately serves both documentation and storytelling and can't be cleanly separated

From the transcript

on one side of what is like documentation so this is the type of video that it could be a personal video where you're taking…

1:06:30

if we are generating fake versions of reality to full people like there's just nothing good about that right and we want to stay away…

1:07:30

on the other side you have what we think of is like storytelling now this could be ads it could be social media posts it…

1:07:30

how do we design products and build products that specifically make it really hard to use on one side and really easy to use on…

1:08:30

From the episode

How to win in the AI era: Ship a feature every week, embrace technical debt, ruthlessly cut scope, and create magic your competitors can't copy

Gaurav Misra (CEO and co-founder of Captions)