Best practices for designing and operating agentic-scale services
1 minute read
About this webinar
Agentic AI services have massive potential—but scaling them remains a challenge. Agents operate at exponentially greater scale, straining infrastructure. LLMs introduce latency and availability issues. Frequent model training leads to constant updates. And most of all, the cost of running LLMs at agentic scale adds up fast.
In this webinar, Tyler Jewell and Richard Li break down how to handle these challenges and build services that scale, including:
Making agentic services fail-proof—even when their LLMs aren’t
Scaling to 2M TPS workloads with a real-world RAG case study
Balancing cost vs. quality tradeoffs when LLMs can cost up to 100,000x more than a database transaction
Redeploying retrain-hungry agents without downtime
[BlogPost 192321515086 Announcing the Akka Agentic Platform, BlogPost 192362253737 Introducing Akka’s new Agent component, BlogPost 192608038701 Video: Overview of Akka Agentic Platform, BlogPost 192605307665 Demo: Build and deploy a multi-agent system with Akka, BlogPost 189650992783 Design patterns for agentic AI, BlogPost 189518647810 New Akka deployment options: elasticity on any infrastructure, BlogPost 189449484665 AI agents are coming for your SaaS stack, BlogPost 189258887023 InfoQ podcast: building agentic AI systems at scale, BlogPost 186949049147 A blueprint for agentic AI services, BlogPost 185530083146 99.9999% app availability, BlogPost 183197704559 Multi-region replicated apps, BlogPost 182678270528 Lightbend is now Akka, BlogPost 180083233087 Akka license keys and a no SPAM promise, BlogPost 180083233324 Is community-backed open source software worth the risk?, BlogPost 180083233603 Open source is at a crossroads, BlogPost 180083233709 Lightbend aims to democratize distributed systems for developers, BlogPost 179858519930 Cloud architecture will fail you. Distributed application architecture will not.]
