Spark is Mosaic's AI assistant. It helps you build and refine journeys, explore your identity and risk data for analytics and fraud investigation, and answer questions about Mosaic features and documentation. Spark uses a combination of Transmit-hosted and third-party large language models, together with Mosaic documentation, to provide contextual, documentation-backed answers in natural language.
Spark is available only to Global Admins and is in active development. Some responses may require refinement. Always validate outputs before using them in production.
We designed Spark to assist you much like a real Mosaic expert. Like AI chat applications, Spark uses previous conversation for maintaining context, so you can ask follow-up questions without repeating all the details. Spark can answer questions about Mosaic features and concepts directly in the chat, grounding its explanations in official documentation where available.
Spark helps you build and understand journeys. With Spark:
- Generate a journey from a short prompt - Spark creates a journey that you can review and adjust in the journey building canvas.
- Understand an existing journey through clear explanations of its intent and logic.
- Ask for guidance on how to structure a journey or achieve a specific outcome by describing it in your own words.
Spark helps you leverage your historical identity and risk data for analytics and decision-making. You can:
- Extract any insights you find useful for your own analytics.
- Ask about changes, comparisons, or anomalies in your metrics.
- Get summaries of key signals relevant to authentication, onboarding, or fraud prevention.
Spark helps you investigate and understand risk and fraud events:
- Break down specific Security Insights to understand their root causes, possible trends, and implications.
- Clarify the reasoning behind a risk recommendation.
Spark also functions as a standalone documentation assistant. Use Spark to find answers about Mosaic features, concepts, and best practices. Ask questions about how features work, implementation approaches, or troubleshooting guidance, and Spark provides answers grounded in official Mosaic documentation.
Spark is designed with security and privacy in mind:
- Spark does not modify your tenant data.
- Information is not stored beyond what is required for session-level analytics and performance monitoring.
- Your conversations and data are processed securely and are not used to train third-party models.
- Spark only accesses data that you explicitly query or that is necessary for the specific task you're performing.