Bot Memories
Bot memories are automatically created during conversations and store information that the bot has encountered or learned. This memory system enables bots to maintain context, recall previous interactions, and provide more personalized and contextually relevant responses over time.
Searching Bot Memories
Searching bot memories enables you to query stored information associated with a specific bot, allowing you to retrieve relevant details from past conversations and interactions. This functionality is essential for understanding what information the bot has learned, auditing conversation history, and debugging bot behavior.
The search helps you discover what the bot knows about specific topics, verify that important information was captured correctly, and analyze patterns in the data the bot has access to. This is particularly valuable for troubleshooting, optimization, and ensuring the bot maintains accurate and appropriate information.
The search operation performs semantic similarity matching against stored memories, returning the most relevant results based on your query. This goes beyond simple keyword matching to understand the meaning and context of your search, finding memories that are conceptually related even if they don't contain the exact search terms.
Each returned memory includes its unique identifier, the text content, and any associated metadata. The results are limited to 50 memories per search to ensure performance and manageability. The most relevant memories appear first in the results, ordered by their similarity to your search query.
Memory search is particularly useful for several scenarios: verifying that the bot correctly learned information from conversations, troubleshooting why a bot might be giving certain responses, analyzing what topics users are discussing with the bot, and auditing the information landscape that influences bot behavior.
The search is scoped to the specific bot you're querying, ensuring that results are relevant to that bot's conversation history and learned information. This isolation prevents cross-contamination between different bots and maintains clear boundaries between bot-specific knowledge.
Note: Memory search only returns memories associated with the specified bot. Memories are automatically created from conversations but can also be manually created or imported to provide the bot with specific knowledge or context.