Building an agent that works once is easy; building an agent that works reliably for thousands of users is an architectural challenge. This session bridges the gap between experimental notebooks and deployed systems, focusing on the specific engineering disciplines needed for success. Join us to learn practical strategies for: 1. System Design: architecting decoupled, scalable agent backends from day one. 2. Continuous Evaluation: moving beyond "vibes-based" testing to metrics-driven evaluation suites that ensure reliability. 3. DevEx & Tooling: streamlining the developer experience to tighten feedback loops and ship improvements faster using open-source frameworks.
What this session is about
Live updates related to this session LIVE
Sourced via Parallel AI Monitor — continuous web watch on 21 topical streams. Updated .
- truefoundry.com Scaling infra for agent workloads
Agent Harness Best Practices: 10 Rules for Production Agents
TrueFoundry materially updated “Agent Harness Best Practices: 10 Rules for Production Agents” and changed its publication date to September 26, 2026. The update adds a DevRev Enterprise-Bench evaluation covering 14 cross-system tasks that require MCP calls across three systems, w
- venturebeat.com Scaling infra for agent workloads
AI agents are breaking the batch-era assumptions behind ...
VentureBeat reported that AI-agent and RAG workloads are breaking batch-era assumptions in enterprise object storage, creating scaling-related failure modes as agent workloads grow. The article argues that storage architecture must evolve to handle the access patterns and load ge
- cloud.google.com Scaling infra for agent workloads
Announcing PostgreSQL for agents in AlloyDB
Google Cloud announced AlloyDB’s agentic database architecture, including an agent pool that can scale from zero to thousands of nodes for bursty agent activity and scale back down when demand falls. This directly addresses database capacity management for highly variable agent w
- sg.finance.yahoo.com Scaling infra for agent workloads
CoreWeave to Offer NVIDIA Vera, the First CPU Built for AI ...
CoreWeave announced that NVIDIA Vera CPU rack-scale systems, designed for demanding agentic AI workloads, would be available on its cloud. The system places 128 CPUs and 11,264 cores in a rack, a new compute-capacity option relevant to scaling agent-native workloads.
- cloud.google.com Scaling infra for agent workloads
Startup Summit 2026 - The Agentic Advantage
A newly published engineering article outlines seven production patterns for surviving 10,000 users on AI applications. Its scaling guidance includes budget-aware rate limiting that considers the user and request type, plus health-aware load balancing that routes requests only to
External links matched to this session via topic relevance. The KB does not endorse third-party content; verify before citing.