Skip to content

PROMPTS.md

Latest commit

 

History

History
146 lines (100 loc) · 8.59 KB

PROMPTS.md

File metadata and controls

146 lines (100 loc) · 8.59 KB

AI Prompts for Building the Virtual AI Assistant

This file contains a series of prompts that can be used to guide an AI developer in building the Virtual AI Assistant project, as specified in the PRD.md. Each prompt corresponds to a phase in the project's roadmap.

Phase 1: Core Brain

Objective: Implement the core brain of the assistant, including the knowledge graph and LLM integration.

Prompt:

You are an expert C++ developer. Your task is to implement the 'Core Brain' of a virtual AI assistant. The project requirements are detailed in the PRD.md file.

Requirements:

  1. Integrate the existing ElephantMemory library: The project already includes a knowledge graph library called ElephantMemory. You need to use this library to store and retrieve information.
  2. Integrate a Large Language Model (LLM): You need to integrate an LLM for natural language understanding and generation. The PRD.md suggests using Ollama or llama.cpp. You should choose one and implement the integration.
  3. Create a CoreBrain class: This class will encapsulate the logic for the thinking process. It should be able to:
    • Receive a text query from the user.
    • Query the ElephantMemory knowledge graph based on the user's query.
    • Formulate a prompt for the LLM, including the context from the knowledge graph.
    • Send the prompt to the LLM and get a response.
    • Process the LLM's response and store new information in the knowledge graph.
    • Return a text response to the user.
  4. Update the CMakeLists.txt: You need to update the build system to include the new CoreBrain class and any new dependencies.
  5. Create an example: You should update examples/main.cpp to demonstrate the functionality of the CoreBrain class. The example should show how to ask a question, get a response, and how the knowledge graph is updated.

Technical Specifications:

  • Programming Language: C++
  • Core Libraries:
    • Knowledge Graph: ElephantMemory (existing)
    • LLM Integration: Ollama or llama.cpp
  • Build System: CMake

You should start by exploring the existing ElephantMemory code to understand how it works. Then, proceed with the LLM integration and the creation of the CoreBrain class.

Phase 2: Perception

Objective: Implement the perception system, including continuous voice monitoring, VAD, speaker diarization, and speech-to-text.

Prompt:

You are an expert C++ developer. Your task is to implement the 'Perception' system of a virtual AI assistant. The project requirements are detailed in the PRD.md file.

Requirements:

  1. Continuous Voice Monitoring: Implement a system that continuously listens to the audio stream from the microphone. The PRD.md suggests using PortAudio or RtAudio.
  2. Voice Activity Detection (VAD): Integrate a VAD library to detect when a user is speaking.
  3. Speaker Diarization: Implement a mechanism to identify when the assistant is being spoken to. This could be a simple keyword-based activation (e.g., 'Hey, Assistant!').
  4. Speech-to-Text: Integrate a speech-to-text library to transcribe spoken language into text. The PRD.md suggests Vosk or a similar library.
  5. Create a Perception class: This class will encapsulate the logic for the perception system. It should be able to:
    • Start and stop the audio monitoring.
    • Detect voice activity.
    • When activated, transcribe the user's speech to text.
    • Pass the transcribed text to the CoreBrain for processing.
  6. Update the CMakeLists.txt: You need to update the build system to include the new Perception class and any new dependencies.
  7. Create an example: You should create a new example file to demonstrate the functionality of the Perception class. The example should show how to start listening, detect speech, and transcribe it to text.

Technical Specifications:

  • Programming Language: C++
  • Audio Processing: PortAudio or RtAudio
  • Speech-to-Text: Vosk or a similar library
  • Build System: CMake

You should start by researching and choosing the appropriate libraries for audio processing and speech-to-text. Then, implement the Perception class and integrate it with the CoreBrain from Phase 1.

Phase 3: Action

Objective: Implement the action system, including TTS and command execution.

Prompt:

You are an expert C++ developer. Your task is to implement the 'Action' system of a virtual AI assistant. The project requirements are detailed in the PRD.md file.

Requirements:

  1. Text-to-Speech (TTS): Integrate a TTS library to convert text responses from the CoreBrain into speech. The PRD.md suggests eSpeak, Flite, or a similar library.
  2. Command Execution: Implement a mechanism to execute shell commands on the computer. This will allow the assistant to perform tasks like opening applications, running scripts, etc.
  3. Create an Action class: This class will encapsulate the logic for the action system. It should be able to:
    • Receive a text response from the CoreBrain.
    • If the response is a command, execute it.
    • If the response is conversational, convert it to speech and play it to the user.
  4. Update the CMakeLists.txt: You need to update the build system to include the new Action class and any new dependencies.
  5. Create an example: You should create a new example file to demonstrate the functionality of the Action class. The example should show how to convert text to speech and how to execute a shell command.

Technical Specifications:

  • Programming Language: C++
  • Text-to-Speech: eSpeak, Flite, or a similar library
  • Build System: CMake

You should start by researching and choosing a suitable TTS library. Then, implement the Action class and integrate it with the CoreBrain.

Phase 4: Integration and Testing

Objective: Integrate all the components and perform thorough testing.

Prompt:

You are an expert C++ developer. Your task is to integrate and test the virtual AI assistant. You need to combine the CoreBrain, Perception, and Action systems into a cohesive application.

Requirements:

  1. Create a main application: Create a main.cpp file that initializes and runs the CoreBrain, Perception, and Action systems. The application should start by listening for the activation keyword.
  2. Integrate the components:
    • The Perception system should pass the transcribed text to the CoreBrain.
    • The CoreBrain should process the text and return a response.
    • The Action system should receive the response from the CoreBrain and either speak it or execute it as a command.
  3. Thorough Testing: You need to test the entire workflow:
    • Test the activation keyword detection.
    • Test the speech-to-text transcription.
    • Test the interaction with the CoreBrain and the knowledge graph.
    • Test the text-to-speech output.
    • Test the command execution.
  4. Update the CMakeLists.txt: You need to update the build system to create the final executable.
  5. Documentation: You should update the README.md with instructions on how to build and run the complete virtual AI assistant.

Technical Specifications:

  • Programming Language: C++
  • Build System: CMake

You should focus on creating a stable and reliable application. Make sure to handle errors and edge cases gracefully.

Phase 5: Deployment

Objective: Package the assistant as a standalone executable and deploy it to users.

Prompt:

You are an expert C++ developer. Your task is to package the virtual AI assistant as a standalone executable for different operating systems.

Requirements:

  1. Cross-Platform Build: Ensure that the project can be built on Windows, macOS, and Linux. You might need to adjust the CMakeLists.txt file to handle platform-specific dependencies and settings.
  2. Create a Standalone Package: Create a script or use a tool (like CPack with CMake) to package the application and all its dependencies into a single, easy-to-install package for each target platform.
  3. Installation Instructions: Create detailed installation instructions for each platform. This should be added to the README.md or a separate INSTALL.md file.
  4. Release on GitHub: Create a new release on the project's GitHub repository with the packaged executables.

Technical Specifications:

  • Build System: CMake
  • Packaging: CPack or similar

You should aim to make the installation process as simple as possible for the end-user.