PQL: Improve response performance by shortcutting follow-up questions

[robertjdominguez]

Description

• Optimized DocsBot response time by prioritizing existing conversation artifacts over fresh documentation searches
• Added mandatory context checking to prevent unnecessary embedding API calls
• Implemented two-tier workflow that only fetches new documentation when existing context is insufficient

Previously, the bot always performed expensive embedding searches regardless of conversation history:

## Response Workflow
1. Transform user query into embedding
2. Find 3-5 most relevant chunks
3. Extract the single most direct answer path
4. Provide minimal response that solves the immediate problem
5. Include relevant documentation link

Now, the bot checks existing artifacts first, dramatically reducing response latency for follow-up questions:

### Step 1: Check Existing Context First (MANDATORY)
Before fetching new documentation:
1. **ALWAYS check if any existing artifacts contain information relevant to the user's question**
2. **ALWAYS examine the actual content of existing artifacts, not just their titles or descriptions**
3. **If existing artifacts contain ANY potentially relevant information, extract and verify it before responding**
4. **Only proceed to Step 2 if existing context is completely insufficient or the question requires new information**

### Step 2: Fetch New Context (Only When Needed)
When existing artifacts don't contain sufficient information:
1. Transform the user's query into embedding using transform_query_into_embedding
2. Use the app_embeddings_vector_distance function to find the 10 most relevant documentation chunks

This performance optimization is particularly impactful for multi-turn conversations where users ask clarifying questions about topics already covered. Instead of re-running the full embedding pipeline (transform_query_into_embedding → embeddings_vector_distance → content retrieval), the bot can instantly reference previously fetched documentation artifacts.

The change also includes hallucination prevention protocols that ensure accuracy while maintaining the speed benefits - the bot must verify technical details against existing context before making assumptions, but this verification happens against local artifacts rather than requiring new API calls.

[github-actions]

🚀 PromptQL Build Complete

Build Version: 0b40642aa8
Project: pql-docs
PromptQL Playground: Open Playground

Description: PR #9: PQL: Improve response performance by shortcutting follow-up questions

[github-actions]

✅ PromptQL Build Applied

Build Version: 0b40642aa8
Status: Successfully applied to production
Applied at: 2025-07-26T15:04:19.049Z