Rapid AI Prompt is the central component for your users to interact with AI based processing. – Yes, the core part still is the text input for the user to enter his/her question – and the streaming text output to show the response. But there is much more:
On the input side, the user can:
- Input text
- Upload attachments: graphics, sound, video
- Talk!
On the output side, the LLM responds in structured text (markdown format). – The LLM processing itself is contacted through some API processing and connects by http-interface to the LLM processing of e.g. ChatGPT, Claude, Ollama or to other agent systems.
Super-important: you can integrate own tools into the prompt processing. Theses tools can be “dark tools” (services) or “UI tools”. Each tool has some API – and is invoked by the LLM processing whenever the description of the tool matches a certain task to execute.
A “UI tool” is directly embedded into the prompt processing. Example:
In the example a graphical tool was added that renders data into a chart. The LLM processing now uses this tool in order to perform typical chart drawings – instead of creating charts as JPG-image on its own. Advantage: the chart will now always look the same. And it provides advanced functions, e.g. tooltip support, reaction on user-selections, download of the chart, etc.
It is very easy to define own tools, both on the “dark side” and on the “UI side”. The only things you need to know: Java as programming language.
The tool drawing charts is only one, quite generic example. You may think about e.g. adding a stock list of available articles and integrate it into the prompt whenever the user should select some article to be delivered.
More technical information is available HERE.
