The Agent Harness Workshop

Опис

Invitation OnlyTrueFoundry Agent Harness: Create Production-Ready AI AgentsAI agents are easy to demonstrate but hard to operate. Once an agent interacts with real users, tools, data, budgets, and internal systems, the challenging issues of routing, permissions, tracking, retries, evaluations, security measures, and deployment become relevant.Join TrueFoundry and HackerSquad for an evening workshop on the "Agent Harness": the control layer around an agent that helps teams connect models and tools, observe every step, manage routes and fallbacks when providers fail, handle costs and quotas, and enforce policies before an agent takes action.TrueFoundry's AI Gateway, Agent Gateway, and MCP Gateway will shape the night. We will focus on the infrastructure patterns teams need as they transition from agent prototypes to production systems: model routing, tool access, MCP workflows, observability, reliable execution, budget controls, and governance.This is a technical build session, not a passive product presentation. Please bring a laptop and think about how your agent would behave under real production conditions.Why This Becomes Difficult in ProductionAn agent harness needs to coordinate models, tools, permissions, storage, context, retries, and fallbacks without turning into a brittle pile of one-off glue code.Production teams need tracking, audit logs, usage visibility, cost allocations, and evaluation cycles to understand what an agent has done, why it did it, and what changed between runs.Scaling agents means managing quotas, budgets, latency, provider failures, tool access, and security policies across teams rather than relying on a single prototype workflow.Security measures must become runtime controls, not just input guidelines: teams need clear rules for which data an agent can see, which tools it can call, which actions require human approval, and how to block, log, or escalate unsafe requests or outputs.Deployment becomes more challenging when agents run within existing infrastructure, connect with internal systems, need to respect governance rules, and must remain reliable as usage grows.Who Should ApplyPlatform, infrastructure, and backend developers building or operating AI agent systemsAI developers and product builders working with LLMs, tool calls, MCP, evaluations, or agent workflowsTechnical founders transitioning agent prototypes to productionSecurity, DevOps, and ML platform practitioners considering governance, observability, and cost controlBuilders looking for hands-on insights into model routing, fallbacks, tracking, quotas, and controlled tool executionWhat We Will CoverHow to think about an agent harness as a production layer around an agentHow model routing, retries, and fallbacks keep AI applications resilientHow access to MCP tools changes the security and observability requirements for agent systemsHow to track the steps of agents, model calls, tool calls, retries, latency, usage, and costsHow quotas, RBAC, policies, and security measures help teams deploy agents with more confidenceSchedule5:00 PM - Doors, Food, and Networking5:30 PM - Welcome from HackerSquad and TrueFoundry5:40 PM - Technical framing from TrueFoundry around the Agent Harness5:45 PM - Hands-on runs, demos, or builder showcase6:30 PM - Build Your Own Agents!7:30 PM - Present Your Agents in ActionBringA laptopAn agent workflow, prototype, or use case you want to secureQuestions about model gateways, MCP tools, observability, cost control, deployment, or agent governanceThe entry is free and approval-based, as seats are limited. Exact venue details in San Francisco and arrival information will be provided to approved participants once the venue is confirmed.When: Tuesday, August 18, 2026 | 5:00 - 8:30 PM PTWhere: AWS Builders Loft - San Francisco, CA