The proposed feature is a Model Recommendation System that provides users with a suggestion button to switch to the ideal model for their current conversation or task, based on Kagi's internal benchmarks. This feature focuses on simplicity and user control, offering recommendations and a one click approach to pick that model.
Purpose:
To help users select the most effective model for their current task, improving accuracy and efficiency.
To reduce the trial-and-error process of manually testing different models.
To provide transparency about which models excel at specific tasks, based on Kagi's internal data.
Impact on Workflows:
- Simplified Decision-Making: Users receive clear, data-driven recommendations for model selection.
- Improved Task Outcomes: By using the best-suited model, users achieve better results with less effort.
- User Control: Users retain full control over model selection, with the option to accept or ignore recommendations.
User Interaction Scenarios
Task Initiation:
- A user starts a conversation or task, such as asking for help debugging Python code.
Kagi analyzes the task and provides a recommendation, such as, "For coding tasks, we recommend Model X."
Suggestion Button:
- The user sees a "Recommended Model" button near the model selection dropdown.
- Clicking the button switches the conversation to the recommended model, updating the context seamlessly.
Example Use Cases
Coding Assistance:
- A user asks for help debugging Python code.
- Kagi recommends a model optimized for programming tasks, such as one with strong contextual understanding of code.
- The user clicks the "Switch to Recommended Model" button to switch to the suggested model.
Creative Writing:
- A user requests help drafting a poem.
- Kagi identifies the task as creative writing and suggests a model known for generating imaginative and stylistic text.
- The user clicks the "Switch to Recommended Model" button to switch to the recommended model.
Data Analysis:
- A user asks for statistical insights or mathematical calculations.
- Kagi recommends a model with advanced analytical capabilities.
- The user clicks the "Switch to Recommended Model" button to switch to the recommended model for accurate and efficient responses.
Integration with Existing Features
Suggestion Button: Add a "Switch to Recommended Model" button near the model selection dropdown.
Task Context Awareness: Use conversation history and task-specific keywords to determine the optimal model.
Comparable Implementations
Grammarly: Suggests writing tones or styles based on the context of the text.
By implementing this suggestion-based feature, Kagi can provide immediate value to users by simplifying model selection and improving task outcomes, while maintaining user control and minimizing disruption to existing workflows.
