How Sarudo Learns
The automatic knowledge extraction pipeline, how learning happens continuously, and how your AI employee improves over time.
Automatic Knowledge Extraction
Your AI employee learns automatically from every interaction. When you share information during a conversation — a client's preference, a project deadline, a pricing decision — the AI extracts key facts and stores them in your knowledge base. This happens in the background through an automated learning pipeline that runs every 30 minutes. You do not need to explicitly tell the AI to remember things, although you can. The system is designed to capture important business information naturally from your conversations.
The learning pipeline runs every 30 minutes, processing recent conversations to extract and store new knowledge. Information is typically available for recall within an hour of being shared.
The Learning Pipeline
The learning pipeline works in three stages. First, conversation analysis identifies statements that contain factual information, decisions, preferences, or instructions. Second, knowledge extraction structures this information into categorized entries with proper context and metadata. Third, embedding generation creates vector representations of each knowledge entry, enabling semantic search later. This pipeline runs automatically and requires no intervention from you.
Continuous Improvement
The more you interact with your AI employee, the smarter it gets. Early on, you may need to provide more context and detail. Over time, the AI accumulates a rich understanding of your business, contacts, preferences, and working style. It remembers that you prefer bullet points in emails, that your biggest client's CEO goes by "Dave" not "David," and that your standard payment terms are net-30. This accumulated knowledge makes every subsequent interaction more efficient.
Using learned knowledge
The AI applies what it has learned.
What Gets Learned
The AI extracts and stores several types of information: contact details and preferences, company information and relationships, business decisions and their rationale, pricing and terms you have quoted, project details and timelines, your personal preferences and working style, standard operating procedures, and industry-specific knowledge. It does not store sensitive information like passwords or financial account numbers. The learning system is designed to capture useful business context, not personal secrets.