The Square Filters Apollo 13 Didn’t Have, and What Incompatibility Almost Cost

Solo founders should record each asset’s source material, AI involvement, human edits, approvals, rights, and published version as the work moves from draft to distribution. Keeping that record beside the asset makes later disclosure a routine publishing step instead of a forensic exercise.

In April 1970, Apollo 13’s crew faced rising carbon dioxide after an oxygen-tank explosion forced them into the lunar module. The command module carried square lithium hydroxide canisters, while the lunar module used round ones. The crew had filters, but the incompatible shapes created a life-threatening problem.

On the ground in Houston, NASA engineer Ed Smylie and his team had to design an adapter using materials already aboard the spacecraft. They worked from a known inventory, assembled a procedure, and sent instructions to the crew. The improvised device worked.

The episode is documented in Jim Lovell and Jeffrey Kluger’s book Lost Moon. Its useful lesson for a solo founder is smaller in scale but similar in mechanism: when requirements change under pressure, an accurate record of the available parts determines whether you can respond quickly.

Treat provenance as part of the asset

Google’s rollout of AI content labels in its asset studio is another sign that disclosure is moving closer to the tools where creative work gets assembled. A label may appear at publication, but the evidence needed to apply it begins much earlier.

For every article, image, social post, newsletter, or video, create a compact provenance record when the first draft is generated. Capture:

• The product and campaign the asset belongs to.
• The source material used, including repository files, approved screenshots, recordings, research links, and product facts.
• Which parts were generated or materially altered with AI.
• Which model or tool produced the initial output, when that information is available.
• The human edits that changed claims, visuals, voice, or meaning.
• The person who approved the final version and the approval date.
• The rights or permissions attached to source media.
• The exact file or revision sent to each destination.

You do not need a sprawling compliance system. A structured record attached to the asset is enough to answer the practical questions later: Where did this come from? What did AI change? Who checked it? Which version went live?

Record decisions while they are cheap to remember

The expensive moment to reconstruct provenance is five minutes before publication. The prompt may be gone. The image could have passed through three tools. A caption might contain a product claim copied from an older page. The approved screenshot may have been replaced during editing.

Record the decision when it happens. When a founder approves an AI-generated outline but rewrites the customer example, note that distinction. When a video uses authentic product recordings with AI-generated narration, preserve both facts. When one social destination receives a shortened edit, connect that derivative to the approved parent asset.

This is the creative equivalent of NASA knowing which materials Apollo 13 actually carried. Smylie’s team could not design around equipment they wished had been aboard. A founder should not make a disclosure decision based on a vague memory of how an asset was produced.

A grounded draft also makes the record easier to maintain. The workflow in How Can I Turn 15 Friday Minutes Into a Grounded Launch Draft? starts with approved inputs, which gives provenance somewhere concrete to begin.

Keep disclosure separate from approval

Disclosure and editorial approval answer different questions.

Approval asks whether the asset is accurate, on-brand, safe to publish, and appropriate for the destination. Disclosure asks how the asset was created and whether that process needs to be communicated through a label, metadata field, caption, or platform control.

Keep both decisions in the record. A founder might approve the substance of a post while holding publication until the correct AI setting is confirmed. That pause should not erase the work or force someone to rebuild its history later.

Marketing Agent can help keep this product-scoped. Its strategy, source-grounded audits, content generation, authentic screenshots and recordings, approval rules, publishing connections, and audit trails sit within the same per-product workspace. Provider access and labeling requirements can still vary, so automation should preserve the evidence and stop when a required decision has not been made.

That boundary matters most for unattended publishing. A scheduled workflow needs a defined rule for unknown disclosure status: pause the asset, retain its provenance, and put the decision in front of a human. Silence should never count as approval.

Build the publication packet now

Start with one reusable publication packet for every asset. Include the final creative, source list, AI contribution summary, rights status, approval record, destination variants, disclosure decision, and publication identifiers returned by supported channels.

Then test it on one real piece of content this week. Take a blog post through draft, human revision, approval, and social adaptation. At each handoff, ask whether another person could determine what happened without opening your browser history or searching an old chat.

Apollo 13’s improvised adapter succeeded because the solution matched the materials on board. Your disclosure process should work the same way. Record the parts while they are in front of you, and the label at the end becomes a controlled publishing choice rather than an emergency reconstruction.