LLMday’s June 19 Bangalore program lists talks on MCP-based production traffic governance, infrastructure-layer observability, and OpenTelemetry tracing for AWS Strands agents on Lambda, reflecting a broader industry focus on managing agentic AI systems in production.
LLMday has published a June 19 Bangalore program focused on the operational challenges of running AI agents in production.
The LLMday Bangalore agenda lists sessions that address what the event describes as a “production gap” in AI systems. According to LLMday, the program includes material on MCP-based multi-agent production traffic governance, infrastructure-layer observability, and OpenTelemetry tracing for AWS Strands agents running on AWS Lambda.
Those topics point to a practical concern for teams moving beyond prototypes: once applications rely on multiple models, tools, and agent-to-agent interactions, organizations need ways to monitor behavior, control traffic, enforce access rules, and debug failures across distributed components.
The governance theme in the LLMday program aligns with recent technical material from Kong Inc. In an engineering post, Kong describes a setup that combines the Strands SDK, Kong AI/MCP Gateway, and Amazon Bedrock to build AI agents with gateway-level controls. Kong says the architecture can support authentication, traffic control, and observability for agentic applications.
Kong has also announced Agent Gateway capabilities intended to govern LLM, MCP, and agent-to-agent communication from a unified control plane. The company frames that work as an extension of AI gateway infrastructure for environments where agents communicate with models, tools, and other agents.
The LLMday listing does not by itself claim a new product launch. Rather, it shows that production governance for multi-agent systems is now a dedicated conference topic, with MCP-related traffic management and infrastructure observability appearing alongside application-level agent development.
Another listed LLMday topic is OpenTelemetry tracing for AWS Strands agents on Lambda. OpenTelemetry is commonly used to collect traces, metrics, and logs across distributed systems. In the context of serverless AI agents, tracing can help operators follow requests as they move through Lambda functions, model calls, tools, and gateways.
The source description specifically mentions AWS Strands agents on Lambda, but it does not provide implementation details in the excerpt. The significance is that observability is being discussed at the infrastructure layer, not only at the prompt or model-output layer.
For production deployments, that distinction matters. Model responses can fail for reasons unrelated to model quality, including authentication errors, rate limits, network failures, tool latency, and unexpected routing between services. Tracing and traffic governance can give engineering teams a clearer view of those failure modes.
The Bangalore program reflects a maturing set of concerns around AI infrastructure. Early agent demonstrations often emphasize capabilities such as tool use, planning, and autonomous task execution. The sessions highlighted by LLMday instead focus on the controls needed to operate those systems under real production traffic.
Based on the LLMday agenda and Kong’s related technical and product materials, the key themes are governance, observability, and operational control. Those are the same categories that have long shaped conventional cloud-native systems, now being adapted for applications that use LLMs, MCP servers, and agent-to-agent communication.
For enterprises evaluating agentic AI, the takeaway is not that any single standard or vendor has solved the production challenge. It is that production readiness is increasingly being framed around measurable infrastructure concerns: who can call which tools, how traffic is routed, what happens when a model or tool fails, and whether teams can trace an agent’s actions end to end.
LLMday has published a June 19 Bangalore program focused on the operational challenges of running AI agents in production.
Production AI agents take center stage The LLMday Bangalore agenda lists sessions that address what the event describes as a “production gap” in AI systems.
According to LLMday, the program includes material on MCP based multi agent production traffic governance, infrastructure layer observability, and OpenTelemetry tracing for AWS Strands agents running on AWS Lambda.
Continue reading
