Where does my API key go?

Your key is held only in the browser tab and sent directly to OpenAI or Anthropic with the request you trigger. It is never stored, logged, or sent to any Gera server. Closing or refreshing the tab clears it. For Anthropic, the request uses the official direct-browser-access header.

Who pays for the API calls?

You do. Each expansion is one real API call billed to your own provider account at your usage rate. The tool itself is free; the only cost is the tokens you consume on your key.

Will the expander add facts that weren't in my bullets?

The system prompt instructs the model to preserve the original meaning and order and not to invent facts. That said, LLMs can still elaborate or fill gaps, so review the output — especially for anything factual or that you will publish — before using it.

Which model should I choose?

For straightforward expansion, the cheaper, faster models (gpt-4o-mini or claude-3-5-haiku) are usually plenty and keep cost down. Step up to a larger model only if you need more nuanced tone-matching or the bullets are highly technical.

Can I control how long the output is?

Yes. The length selector tells the model to expand each bullet into one sentence, two to three sentences, or a full paragraph. Pick the shortest option that reads well — it costs fewer output tokens and stays closer to your original intent.

Bullet Expander (BYO-key)

LLMs love to answer in terse bullets, but reports, emails, and docs often need prose. This tool turns a bulleted list into flowing text using your own OpenAI or Anthropic key — one bullet in, one expanded item out, in the same order, with the tone preserved.

How it works

Pick a provider and model, paste your API key, and drop in your bullets. Choose how far each point should be expanded — a single sentence, two or three sentences, or a full paragraph. When you run it, the tool makes one direct request from your browser to the provider with a system prompt that tells the model to keep the original meaning and order and not to fabricate detail. The expanded text comes back ready to copy.

Your key never touches a Gera server: it lives only in the tab and is sent straight to OpenAI or Anthropic (using the official direct-browser-access header for Anthropic). Refreshing the page clears it.

Getting good results

Match the length to the medium. “One sentence” is great for turning meeting notes into a readable summary; “paragraph” suits a draft you’ll edit further.
Keep bullets self-contained. The model expands each line independently, so a bullet that depends on context from another may lose nuance — add that context inline.
Review factual claims. Expansion is generative; the model can smooth over gaps with plausible-sounding detail. Verify anything you’ll publish.

Tips

Pair it with the Text → Bullets Collapser to round-trip between outline and prose as you draft.
Cheaper models handle expansion well — save the premium models for nuanced or technical copy.
Cost scales with output length, so the shortest length that reads well is also the cheapest.