Creating a Multi-Tenant AI Agent Platform Handling 7K+ Sessions Without Cross-Team Interference
Summary
A highly scalable multi-tenant AI agent platform called BYOP enables Salesforce teams to independently build, deploy, and scale custom reasoning engines without the interference issues of a monolithic system. It addresses the challenges of cross-team code coupling, resource contention, and bottlenecks in deployment processes by isolating each agent with dedicated infrastructure and self-service pipelines. The platform supports complex multi-turn workflows with strict tenant isolation and robust session management, validated by production use-cases handling thousands of concurrent AI sessions with low latency. Salesforce teams can leverage BYOP to accelerate AI-driven workflow innovation while maintaining system stability and autonomy.
Takeaways
- Isolate AI reasoning engines to avoid cross-team interference and resource contention.
- Use self-service CI/CD pipelines to enable independent agent deployment and rapid iteration.
- Leverage platform-provided session management and streaming to simplify reasoning logic development.
- Implement strict tenant isolation by embedding tenant and session IDs in storage keys with TTL policies.
- Adopt distributed tracing and cost attribution mechanisms for observability and performance tuning at scale.
In our Engineering Energizers Q&A series, we highlight the engineering minds driving innovation across Salesforce. Today, we spotlight Priyanka Saraf, Senior Software Engineer on the Agentforce Foundations team. Priyanka is the engineer behind Bring Your Own Planner (BYOP), a multi-tenant AI agent platform that enables teams to develop, deploy, and scale independent custom reasoning engines on shared infrastructure. The system now supports over 7,000 active agent sessions and 14,000–15,000 daily requests across production environments. Explore how the team eliminated cross-agent interference caused by a monolithic planner where shared infrastructure and deployment pipelines slowed development across more than 100 engineers — and how the team scaled multi-tenant AI agents on Agentforce while maintaining strict isolation and platform overhead as low as 5 milliseconds per request.