===================== Current Behavior: ===================== When asking Hatz to refresh or regenerate a document (e.g., rewriting or updating a Word document, report, or structured output), the results frequently require significant follow-up prompting, corrections, and coaxing to reach an acceptable quality level. This is especially noticeable when compared to performing the same task directly in Claude, where the output closely matches the structure, tone, formatting, and content of the original document on the first attempt. The gap appears to stem from how the underlying document generation pipeline handles templating — things like preserving heading hierarchy, table structure, inline formatting, section relationships, and stylistic consistency are often lost or inconsistently applied when Hatz processes the request. ===================== Desired Behavior ===================== Hatz's document generation and refresh capabilities should produce output quality comparable to what Claude delivers natively when given the same prompt and source material. This includes accurate preservation of document structure, formatting conventions, and content relationships without requiring multiple follow-up corrections from the user. The templating engine or generation pipeline should be capable enough that a single well-formed prompt yields a usable, high-quality result. ===================== Use Case ===================== When refreshing existing documents — such as updating a proposal, regenerating a report, or rewriting a structured business document — users should be able to trust that Hatz will handle the task in one pass without needing to re-prompt repeatedly or manually fix formatting and structural issues afterward. This is a major client concern that comes up in nearly every deployment. It is not an edge case — it is one of the most consistently asked questions during onboarding and adoption, and it creates an immediate credibility gap when clients compare the results side by side with Claude directly. If Hatz is positioned as a business productivity tool, document generation quality needs to be reliable enough for professional use without extra overhead. The current experience creates friction and reduces confidence in the platform, and addressing this would have a direct and measurable impact on client satisfaction across deployments.
