In early 2023, we started our very own Prompt Command Center. It has become a very useful tool throughout our systems.
A “Prompt Command Center” is a central repository of prompts, to which you attach more and more sophisticated functionality as it becomes available.
Our “Prompt Command Center” also contains a prompt library, an AI Agents library, and a prompt chaining tool.
At the core is the Edit Prompt control panel, which allows easy selection of attributes.
Model settings: This allows you to select which model you want to use ranging from OpenAI, Azure Open AI, or Anthropic with its large context. New LLM model can be added for ease of updating your AI.
This also allows you to set preset behaviors, such as predictability and creativity, optimized for certain use cases.
Actions/Function calling: This allows you to select which actions the AI is allowed to expose to the users, allowing fine tuning and secure action ability in each case.
Meta data details: This is an AI summary of what that particular prompt does. So the AI summarizes the use of the agent, so that Multiple Agent Systems can then find the best sub-agent to use.
Additionally, there are controls for advanced settings and expected output.
Having your own Prompt Command Center allows you to collect and centralize AI intelligence inside your organization today, for use by smarter AI systems tomorrow.
This can then be used easily in the following scenarios:
- Equivalent to CustomGPT, as isolated AI agents.
- AI Email agents – agents that have access to your internal data.
- MAS – multiple agent systems – team of experts who converse and make better decisions.
- “AI functions” – here we can call them in code, so use an LLM prompt in place of a logically coded function(), when appropriate.
If you are wondering where to start with AI internally, this is one way forward.
If you need any help or tips please reach out, I am always happy to spread knowledge learnt on this AI journey.
Jobe.ai 
Leave a comment