Refining Content and Working with History

History: your content library

Every piece StretchBAS produces — single generations, each piece of a content pack, every social post, and every refinement — is saved to History. History is your working library: browse it, copy content into your channels, score it against a fingerprint, or refine it further. Nothing you generate is ever lost, and each entry records its content type, the request that produced it, the model used, the credits spent, and when it was created.

Open it from StretchBAS → History. Entries are listed newest-first, and you can filter to a specific brand profile.

What refining does

Refine rewrites an existing generation according to a plain-English instruction — "make it punchier," "add a stronger call to action," "shorten to two sentences," "make it warmer" — while keeping it on-brand. StretchBAS loads the original content and its brand fingerprint, applies your instruction, and saves the result as a new generation (the original is preserved). Each refine spends 1 credit and is refunded if it fails.

Because refinements are saved as new entries (their request is stamped refine: …), you build up a clear trail of every version — handy for A/B choices and for showing a client the iteration.

Refining step by step

1. Open StretchBAS → History and select the generation you want to improve.
2. Choose Refine.
3. Write your instruction — describe the change you want in plain language.
4. Run it. The revised version appears as a new History entry alongside the original. This spends 1 credit.
5. Repeat as needed — you can refine a refinement to iterate toward exactly the wording you want.

Example

Beau generates a Google Business post for the Payne Coaching launch, but the first draft buries the offer. He selects it in History and refines with: "Lead with 'Only 8 spots' and end with a clear 'Book your free intro call' CTA." StretchBAS returns a tightened version that opens with the scarcity hook and closes with the CTA — still in Payne's warm voice — saved as a new entry. He keeps both versions and posts the refined one.

The generate → score → refine loop

For high-stakes content, chain the three tools:

1. Generate the first draft from the profile.
2. Score it 0–100 against the fingerprint to see what's on- and off-brand.
3. Refine using the score's suggestions as your instruction (for example "warm up the tone and drop the corporate jargon the score flagged").
4. Re-score to confirm the number went up, and publish.

This loop turns a rough first draft into a confidently on-brand final piece in a few credits.

Refining still works even if the profile is gone

Refine is resilient: it uses the original content plus its fingerprint, but if the source profile has since been archived, refine still works from the prior text and your instruction. You won't be blocked from improving an old piece just because you cleaned up your Profiles list.

Tips

• One change per refine. Clear, single-purpose instructions ("add a CTA") produce cleaner results than a paragraph of mixed requests.
• Use score suggestions as refine instructions — it's the fastest way to fix an off-brand draft.
• Keep versions. Because each refine is a new entry, you can always go back to an earlier draft you liked.
• Filter History by profile when you manage multiple brands so you're always working with the right client's content.

Troubleshooting

• "An instruction is required." You didn't say how to refine. Describe the change you want.
• "Generation not found." The entry isn't in your current workspace, or the id is wrong. Open the piece from History and refine from there.
• The refinement drifted off-brand. The source fingerprint may need correcting — edit the profile's voice/tone/themes, then refine again. If the profile was archived, consider re-absorbing the brand for a stronger result.