AI Text Copilot

Created: 2026-07-31 10:00 · Updated: 2026-07-31 10:00

The Text Copilot is an AI chat assistant right inside the PPPTools Task Pane that helps you rework slide text: shorten it, make it more concise, rephrase it, adjust the tone (more formal or more casual) or turn running text into bullet points. You select one or more shapes on the slide — their text is passed along automatically as context — and describe in the chat what you want done. The result can be written back into the selected shape with a single click.

Opening the Panel

PPPTools Task Pane → ≡ (Hamburger Menu) → 🤖 Text-Copilot

Choosing an AI Engine

The Text Copilot uses the same AI providers as Translate — with the same per-provider API keys. So a Claude or OpenAI key you already stored for translation does not need to be entered again.

You can, however, pick a separate optimisation engine, independent of the translation engine. That way you can, for example, translate with DeepL but optimise with a language model.

For details on creating an API key (including the note that a chat subscription does not include API credit), see the Translate page.

Setting up Engine and Key

1. Open PPPTools Task Pane → ≡ → 🤖 Text-Copilot.
2. Pick a provider in the "KI-Engine" (AI engine) dropdown. A short provider note appears below it.
3. The API key is pre-filled automatically if it is already known from Translate. Otherwise paste the key into the field.
4. Click "Check" — on success a green confirmation appears (a short test chat sent to the model).
5. Click "Speichern" (Save) — the optimisation engine and the key are stored permanently.

Working with the Copilot

1. Select the context

Select one or more shapes with text on the slide. Their text is read fresh on every send and passed to the model as context. The context line at the top shows how many characters were loaded; "Kontext aktualisieren" (Refresh context) updates it manually.

2. Write an instruction

Type freely in the input box what the copilot should do — for example:

• "Shorten this to 3 bullet points."
• "Make this more concise."
• "Rephrase this more formally."
• "Turn this into running text."

Enter sends the message, Shift+Enter inserts a line break.

3. Multiple turns (multi-turn)

The copilot remembers the ongoing conversation. You can follow up — "a bit shorter", "now in English", "drop bullet 2" — and refine the result step by step.

4. Apply to the slide

When the last answer fits, click "In Auswahl übernehmen" (Apply to selection). The text of the selected shape is completely replaced with the answer.

Tips

• Engine choice: For shortening, rephrasing and judging text, AI language models (Claude, OpenAI, Gemini, Mistral) are clearly stronger than pure translators.
• Separating translation and optimisation: DeepL is excellent for translation but weak at shortening/conciseness — hence the separate optimisation engine.
• Keep the original: Applying replaces the entire text. To keep the original, copy the shape first or use Ctrl+Z.
• Watch the cost: For OpenAI/Claude you can set a spending limit in the respective portal.

Error Messages