Memory Episode Access for Troubleshooting

Question: Can agents reference the GPT-5.1-chat API parameters fix episode to troubleshoot chat and voice failures?

Answer: ✅ Yes, agents automatically search and reference this episode during conversations.

How It Works

1. Automatic Memory Retrieval (RAG)

When users interact with agents (Elena, Marcus, Sage), the system automatically:

  1. Extracts the user’s query from their message
  2. Searches memory for relevant episodes, documentation, and past conversations
  3. Injects retrieved context into the agent’s prompt
  4. Agents can reference this context when responding

Code Location: backend/agents/base.py → _reason_node() method

# Automatic memory retrieval (RAG)
memory_results = await memory_client.search_memory(
    session_id="global-search",  # Searches across ALL sessions
    query=query,                  # User's message
    limit=5                       # Top 5 relevant results
)

2. Search Capabilities

The search_memory method uses HYBRID SEARCH combining:

  1. Semantic Search (pgvector) - Vector embeddings for meaning-based similarity
  2. Keyword Search - Full-text matching in session content and metadata
  3. Metadata Match - Title, topics, summary matching

Search Prioritization:

  • Wiki pages (doc-wiki-*) - Highest priority
  • Documentation (doc-*, capability-*) - High priority
  • Canonical sessions (sess-*) - Medium priority
  • Other sessions - Lower priority

3. Episode We Ingested

The GPT-5.1-chat API parameters fix episode was ingested with:

  • Session ID: capability-gpt-5.1-chat-api-parameters-fix-2025-12-31
  • Topics: gpt-5.1-chat, api-parameters, max_completion_tokens, temperature, azure-ai-foundry, chat-endpoint, llm-api, troubleshooting, model-compatibility, api-version, 2024-12-01-preview
  • Metadata: Includes title, summary, and full documentation content

Example Scenarios

Scenario 1: User Reports Chat Failure

User: “Chat is not working, I’m getting error messages”

Agent (Elena) will:

  1. Search memory for: "chat endpoint failing" or "chat error"
  2. Find the episode: capability-gpt-5.1-chat-api-parameters-fix-2025-12-31
  3. Reference the fix in response:
    • “Based on our troubleshooting history, chat failures are often caused by incorrect API parameters for the gpt-5.1-chat model. The model requires max_completion_tokens instead of max_tokens, and doesn’t support custom temperature values…”

Scenario 2: User Asks About Model Configuration

User: “How do I configure gpt-5.1-chat?”

Agent will:

  1. Search memory for: "gpt-5.1-chat configuration" or "gpt-5.1-chat API parameters"
  2. Find the episode with configuration details
  3. Provide accurate configuration guidance from the episode

Scenario 3: Voice Failure Similar to Chat

User: “Voice is not working”

Agent will:

  1. Search memory for: "voice endpoint" or "LLM API error"
  2. Find related troubleshooting episodes
  3. Suggest checking similar issues from chat troubleshooting

Verification

To verify the episode is accessible, agents can use the search_memory tool:

# In agent code
results = await memory_client.search_memory(
    session_id="global-search",
    query="gpt-5.1-chat API parameters max_completion_tokens",
    limit=5
)

The episode should appear in results because:

  • ✅ Session metadata includes relevant topics
  • ✅ Content includes keywords like “max_completion_tokens”, “temperature”, “gpt-5.1-chat”
  • ✅ Session ID starts with “capability-“ which is prioritized in search

Limitations

Important Note: As an AI assistant in this chat interface, I cannot directly search Zep memory in real-time. However:

  1. ✅ Agents (Elena, Marcus, Sage) CAN search and reference the episode
  2. ✅ The system automatically retrieves relevant episodes during conversations
  3. ✅ Users will see agents referencing this troubleshooting knowledge
  4. ✅ The episode is searchable via semantic, keyword, and metadata search

How to Test

  1. Send a chat message like: “Chat is failing, what could be wrong?”
  2. Agent should search memory and find the episode
  3. Agent should reference the troubleshooting steps from the episode
  4. Check agent logs to see memory retrieval in action
  • docs/troubleshooting/gpt-5.1-chat-api-parameters-fix.md - Full troubleshooting guide
  • backend/agents/base.py - Agent memory retrieval implementation
  • backend/memory/client.py - Memory search implementation
  • scripts/ingest-gpt-5.1-chat-api-parameters-fix.py - Episode ingestion script