Conviva Digital Product Insights (DPI) enables the continuous evaluation and optimization of agent conversation performance and outcomes turning every agentic conversation into outcome-driven intelligence that connects user prompts/patterns, conversational sentiment, response patterns and user follow-on actions with optimized business outcomes.
Data Ingest
The DPI sensor ingest agents conversation data as part of the DPI application data ingest. The DPI sensor applies a lightweight integration to your application platforms, enabling automatic, semantic-less data collection across different data sources and formats, with heavy computation done on the server-side. For more details, see DPI Data Ingest.
Outcome Labels and Dimensional Data
As Conviva DPI analyzes agent activity, application use, and user conversations, it builds outcome labels and dimensional data as part of the DPI feature set for advanced analysis and predictive insights. These metrics and metadata enable drill-downs to pinpoint the dimensional cohorts for leading to the highlighted growth opportunties and sources of friction.
Agent dimensions for analysis of active user across agent, chat, and user sentiment
Common Use Cases
The most common DPI agentic use cases involve using Flows based on critical user journeys and drill-downs across agent-related dimensional metrics to optimize system prompts, tool instructions, tool configurations, model selection, model training, UX designs, and more.
Use Case Examples
In this case, the data shows that the Agent is not handling price constraints properly, causing Discover to Add to Cart users to stop before adding to cart. The agent actions underperformed the search to cart approach. The dimensional data demonstrate the agent was not handling price constraints properly, leading users to stop before adding items to the cart.
Launching Agentic Analysis
Access agentic features in the Trends or Trends Overview Dashboard,
Note: .
Conversation Conversation Agent Agent Agentic Chat