During AI Engineer World’s Fair week in San Francisco, Superagentic AI hosted Harness Engineering: State of the Art in Agent Harnesses at AWS Builder Loft. The event received more than 1,000 registrations and welcomed approximately 280+ attendees in person at AWS Builder Loft in downtown San Francisco. The response from the community significantly exceeded expectations. ai queued outside the venue before the event started and the room quickly reached maximum capacity. Throughout the evening, attendees watched talks from hallways, doorways, and every available space inside the venue. For a topic that only a handful of companies have publicly discussed so far, the level of interest demonstrated that the industry is actively looking for deeper conversations around production AI systems and the engineering challenges behind them.
Event Highlights
A short highlights video from the event is available below:
Why Harness Engineering?
Over the past two years the industry has moved from Prompt Engineering to Context Engineering and increasingly towards production AI agents. As agents move into production environments, a new layer of infrastructure is emerging around them.
This layer includes Evaluation, Observability, Memory, Optimization, Experimentation, Reliability, Runtime orchestration, Guardrails, Human feedback loops. Collectively, we refer to this layer as the Agent Harness. The event explored the current state of the art from multiple perspectives rather than promoting a single framework or implementation approach.
Speakers
The event featured speakers from organizations actively working on different aspects of the agent stack:
- Dat Ngo (AI Architect, Arize AI)
- Arun Kumar : CTO and Cofounder of RapidFire AI
- Myeongsoo Kim : Applied Scientist @ AWS AI Labs (Kiro)
- Linghua Jin Cofounder & CEO @ CocoIndex
The feedback from attendees was overwhelmingly positive and many participants requested access to the presentation slides following the event.
Community Support
Superagentic AI would like to thank AWS Builder Loft for hosting the event and supporting the San Francisco AI engineering community. Special thanks also go to Arize AI and CocoIndex for sponsoring food and refreshments for attendees. Most importantly, thank you to the hundreds of engineers, founders, researchers, and practitioners who attended and contributed to the discussions throughout the evening.
Agent Engineering HQ
Harness Engineering was the first event under the new Agent Engineering HQ initiative, community-first forum where practitioners can discuss the technical challenges involved in building production AI systems. The intention is to focus on practical engineering lessons, implementation details, failures, trade-offs, and real-world experience rather than product marketing or vendor positioning. Harness Engineering is only one part of a much larger landscape.
Future events are expected to cover areas including:
- Memory Engineering
- Context Engineering
- Inference Engineering
- Eval Engineering
- Loop Engineering
- *. Engineering
The industry is still defining these disciplines and there is significant value in creating spaces where practitioners can learn from one another. Checkout more at the Agent Engineering events and conferences here
About Superagentic AI
Superagentic AI develops low-level infrastructure and developer tooling for production AI agents. The company recently introduced SuperQode, a Harness Engineering framework designed for coding agents and advanced agent workflows. The framework focuses on allowing teams to define, evaluate, optimize, and evolve their own agent harnesses rather than relying entirely on fixed abstractions provided by external platforms.
Looking Ahead
The enthusiasm from the San Francisco AI community demonstrated that there is significant interest in around the emerging disciplines of Agent Engineering. Superagentic AI plans to continue supporting these conversations through future Agent Engineering HQ events in San Francisco and London. If you are building in this space and would like to speak, sponsor, collaborate, or contribute to future events, we would be happy to hear from you.
