Task Builder — Intelligent Matrix-Based Prompt Generation

September 7, 2025

The Task Builder’s prompt generation engine has been completely rebuilt. Previously, every prompt used the same five-section structure regardless of whether you were generating a simple checklist or a complex multi-output project plan. Now, the system evaluates both your complexity level and how many outputs you’ve selected, then chooses from five distinct templates optimized for that combination. The result: prompts that actually match what you’re trying to accomplish.

Improvements

• Matrix-based prompt generation — The Task Builder now selects from five intelligent templates based on the combination of your complexity level (1–5) and output count. A simple task with a few outputs gets a lean, direct three-section prompt. A complex task with many outputs gets a systematic eight-section framework with output coordination. This covers 75+ distinct complexity/output combinations, each producing an appropriately structured prompt.
• Task type detection — The generation engine now analyzes your task description to detect the type of work (learning, communication, organization, or creative) and adapts the prompt structure to better suit that category.
• Output format intelligence — Selected output formats are now categorized (summary, list, communication, processing) so the generated prompt includes format-appropriate guidance for each output type rather than generic instructions.
• Improved output format descriptions — The output format selection interface now includes clearer descriptions so you know exactly what each format produces before selecting it.
• Updated Task Builder navigation — The Task Builder page header now matches the Creation Hub design with consistent icon styling, clickable navigation back to the hub, and proper spacing between sections.

Infrastructure

• Modular template architecture — The prompt generation system was restructured into a modular template architecture, making it straightforward to add new templates or adjust selection logic as the platform evolves. This improves long-term maintainability without any visible change to users.
• Code quality cleanup — Removed unused code across model selector components and the generation engine, reducing complexity and improving build performance.