High-Level Overview

A Smart Package is a powerful feature that makes your AI chat experience smarter and more focused. Instead of blindly searching through all your documents, it intelligently zeroes in on the most relevant parts — even if the content is spread across multiple sections.

Think of it like "multi-hop reasoning": the system takes your question, finds clues, follows those leads, and builds a custom package with only the information you care about.

Whether your question is broad or highly specific, Smart Package helps you get faster and more accurate answers - with less irrelevant noise.

How to use the Smart Package feature

• At the bottom of the chat interface window is a check box to "Enable Smart Package"
• "Enter specific keywords to search for" - An example could be product names or model numbers. Use the exact description as is in the documentation. For example if you are searching related to Ford pickups. Search for "F150, F250, F350" rather than like "F-150, F 250, Ford350" as those are not accurate keywords. It is ok to list multiple terms with commas between them.
• "Find these words near those keywords" - An example could be "towing, tow rating, load, tow". AiWerkz will find similar words.
• The AiWerkz will generate a custom chat package based on the requested information.
• Paste that chat package into ChatGPT, Google Gemini or whichever Ai system you prefer to use.

Detailed Explanation (For Advanced Users)

1. Initial Signal Extraction:
   Uses your user_query, optional user_query_keywords, and user_query_nearby terms to identify promising chunks via keyword matching.
2. First-Hop Filtering:
   Matches are collected as document_section_id:chunk_id pairs (e.g., abc123_5, def456_9), forming the "first hop".
3. Second-Hop Vector Refinement:
   These first-hop chunks are passed to a vector similarity engine, which finds nearby content using semantic embeddings - even if the wording is different.
4. Final Chat Package Assembly:
   All relevant chunks are pulled from the database and compiled into a clean chat package, along with your query, ready to be sent to the LLM.

Why use Smart Package?

• Enables cross-document context reasoning
• Reduces irrelevant content
• Scales to large document sets
• Great for technical, research, or support queries