Engineering self-improving AI systems
Please also register through the official AWS link here, as AWS Builder Loft requires it for entry. Without AWS registration, guests may not be allowed in.
Overview
AI agents are no longer a side experiment inside AI teams. They are becoming part of product surfaces, internal operations, customer workflows, developer tools, and decision-making systems.
The question now is not just how to build an agent.
It is how to build the harness around it: the systems, workflows, boundaries, infrastructure, and human decisions that decide what the agent can do, where it should stop, how it should recover, and how it gets better after real use.
This room brings together builders working on that layer.
Session Description
We’ll spend the evening with engineers, founders, AI teams, cloud builders, and product teams who are designing what sits around agents once they become part of real systems.
We’ll talk about closing the loop, what comes after the first working agent, how teams are thinking about agent systems in practice, and what builders are learning from real deployments.
Agenda
5:00 PM — Check-in, food, and networking
5:30 PM — Welcome and room framing
5:40 PM — AWS technical segment on agentic AI and Amazon Bedrock
6:00 PM — Closing the Loop: Engineering Self-Improving AI Systems
Nikhil Pareek, Founder of Future AGI
6:30 PM — Self-Improving Retrieval Loops for Reliable AI Agents - John, DevRel at Qdrant
Most agents trust the first search result and keep going, even when retrieval fails. This talk shows how to build a self-improving retrieval loop that detects weak results, changes its strategy, and knows when to retry, escalate, or stop before hallucinating.
6:45 PM — TBA
If interested in speaking/sponsoring, reach out to Vrinda
7:00 PM — Builder demos
5-minute demos from teams building agents, workflows, devtools, infrastructure, voice systems, or applied AI products. Interested in pitching? Submit here
8:00 PM — Open networking
8:30 PM — Close
Building something relevant to AI agents, workflows, infrastructure, voice, devtools, or applied AI?
We’re opening a few spots for guest speakers, 5-minute builder demos, and aligned sponsors. For any queries, feel free to reach out to here: https://www.linkedin.com/in/vrinda-damani/
Note: All guests must be 18+ and present a valid government-issued physical photo ID at check-in (digital IDs will not be accepted). Scooters and bikes are not permitted in any Amazon building with no on-site parking available.
About Future AGI
Future AGI is an open-source end-to-end agent engineering and optimization platform that makes AI agents self-improve. It gives teams one place to simulate edge cases before launch, evaluate what actually happens in production, protect users from unsafe outputs in real time, and turn every trace into signal for the next version. Everything runs on one platform and one feedback loop, from first prototype to production.