A practitioner-focused webinar covering agent harness engineering — the infrastructure around LLMs that determines agent quality beyond the model itself. Presenters from Temporal and Grid Dynamics define the harness concept, demonstrate durable execution using Temporal workflows, and show a real-world demo combining ReAct and CodeAct patterns to generate a marketing HTML page by scraping Temporal's customer case studies. Key topics include context engineering, guardrails, agent skills, the hybrid ReAct+CodeAct approach, and how Temporal's durable execution handles failures, long-running workflows, and human-in-the-loop approval scenarios. Temporal's 'continue as new' primitive and large payload support for AI workloads are also discussed.
•53m watch time
23 Impressions