Autoresearch in a Multi-Agent AI Village
Project Paradox is an existing multi-agent framework built at Supercell's first AI Innovation Lab, which has a 3D Unity village with local LLM powered agents. The characters remember conversations, update emotional state, track trust, plan actions, move through rooms, transfer items, and talk to each other through a FastAPI backend. The new work is an autoresearch layer around that village. We built a backend loop that runs controlled social scenarios, scores the resulting NPC behavior, proposes protocol or policy changes, reruns the suite, and keeps changes that improve the agents. The goal is to move beyond one good chat response and measure whether an NPC society can preserve source attribution, verify claims, spread important information, coordinate goals, and replan after new information arrives. The talk walks through the system architecture and the lessons from building it. We show the backend simulation harness that executes Unity style actions without opening Unity, the scenario suites that test information diffusion and memory provenance, and the ratchet loop that edits protocol text or planner policy with rollback. One accepted run improved information diffusion by teaching agents to broadcast important sourced evidence while preserving who said it. The practical takeaway is a reusable pattern for AI engineers building agents with messy state. Freeze the harness, expose a small editable policy surface, score real behavior instead of vibes, and let an agent search for improvements under rollback. The same pattern applies to game agents, coding agents, support agents, personal agents, and other systems where long horizon behavior matters more than a single response. Speakers: Erina Karati — Former Microsoft, Supercell; Arunachalam Manikandan — University of Minnesota.
About the Autoresearch Track
Autoresearch sessions at AI Engineer World's Fair 2026 in San Francisco.
When
Wednesday, July 1, 2026
3:45 PM - 4:05 PM·20m
Where
Main Stage
Capacity: 4000 attendees
Sign in to add this talk to your schedule.
Speakers (2)
Erina Karati
AI Engineer
Former Microsoft, Supercell
Erina Karati builds applied AI systems focused on Gen AI, multi-agent architectures, and production-ready ML pipelines, backed by $10,000 in grant-funded AI work, real-world AI innovation at Supercell, and 3 years at Microsoft. At Microsoft, she worked on large, customer-facing production systems in complex enterprise environments. Her work spanned networking, system reliability, security, and debugging distributed failures at global scale, shaping how she approaches robustness and observability in AI systems today. More recently, she worked as an AI Engineer at Supercell, building modular multi-agent systems and scalable AI infrastructure for real-world use cases. She is also the co-founder of MinneDigest, an AI-powered news and podcast platform that won the AI x Journalism Hackathon and secured $10,000 in funding. She graduated with a Master’s in Computer Science from the University of Minnesota with a 4.0 GPA in May 2026, and is especially interested in combining strong engineering foundations with advanced AI to solve meaningful, real-world problems.
