Pismo has a lot of potential, I use it dozens of times per day!
But to be the ideal tool, a few key components are lacking.
A.
Having (at least a brief) history for a follow-up call on the outcome of the last one.
Examples:
1. prompt: Correct and enhance a marked text. Change it out.
2. prompt: Translate it into a language. Add below.
or
1. Correct and improve a marked text.
2. Make it longer or shorter.
3. Rephrase it in any manner we like.
4. Replace the original in its position with the result.
or
simply regenerate the result and let us select the preferred one by comparing it to the original one.

B.
The ability to choose different LLMs for various use cases is essential.
As best to be able to select an LLM for each separate prompt.
An excellent starting point would be BYOK and the access to local models over Ollama integration.