Prompts

TextQL’s prompt system combines state-of-the-art language models with sophisticated context management to deliver precise, customizable AI interactions. Our system leverages multiple leading AI providers and maintains context through a DAG-based architecture, ensuring consistent and accurate responses across complex analytical workflows.

​

AI Model Integration

TextQL integrates with industry-leading language models to provide robust and reliable AI capabilities. Our system supports OpenAI’s GPT-4 and GPT-3.5 series models alongside Anthropic’s Claude models, allowing organizations to leverage the most appropriate AI technology for their specific needs. Model selection and configuration are fully customizable through environment variables, enabling seamless adaptation to evolving requirements.

For semantic understanding and search capabilities, TextQL employs advanced embedding models including OpenAI’s text-embedding-3-small and text-embedding-ada-002, BAAI’s bge-small-en-v1.5, and Sentence Transformers’ all-MiniLM-L6-v2. These models power our sophisticated data exploration and analysis features, enabling natural language understanding across diverse datasets.

​

Core Features

TextQL’s prompt system maintains conversation context through our DAG-based architecture, ensuring AI responses remain relevant and coherent throughout complex analytical workflows. This architecture enables sophisticated state management and context preservation, allowing for multi-step analysis while maintaining accuracy and consistency.

Our semantic search capabilities leverage vector embeddings to enable powerful data exploration across your entire data ecosystem. The system supports multiple embedding models with configurable dimensions and similarity parameters, allowing for precise tuning of search behavior to match your specific use cases.

The system integrates seamlessly with our mode-based processing system, Dakota, which manages different analytical contexts and workflows. For detailed information about our sophisticated mode-based processing system, see our Dakota Modes documentation.

​

Configuration Options

TextQL’s prompt system provides extensive configuration options to meet specific organizational requirements while maintaining security and performance. Model configuration options include provider selection, temperature and creativity settings, token limits, and response parameters. These settings can be adjusted at both the system and mode level, allowing for fine-grained control over AI behavior.

Embedding configuration options enable organizations to select their preferred embedding models, configure vector dimensions, and adjust similarity search parameters. These options ensure optimal performance for different types of data and use cases while maintaining efficient resource utilization.

​

AI Features

TextQL’s prompt system powers several sophisticated AI features that enable natural language interaction with your data. The Object Explorer converts natural language queries into optimized database operations, allowing users to explore data warehouses using conversational language. For example, when asked about “last month’s revenue by region,” the system automatically generates and executes appropriate queries while maintaining context for follow-up questions.

The Metric Explorer provides context-aware metric analysis, understanding complex questions about business metrics and their relationships. When investigating changes in metrics like customer satisfaction, the system considers historical trends, related metrics, and potential causal factors to provide comprehensive insights.

Our Meta Explorer facilitates smart dataset discovery, matching user questions with relevant data sources across your organization. When users need to analyze patterns like user engagement, the system identifies and suggests appropriate datasets while explaining their relevance to the specific analysis needs.