001 / RESEARCH OVERVIEW
Building as a system
Can shared memory and verifiable workflows expand what individuals and small teams can build?
Read the brief ↗RESEARCH LIBRARY / NOAERTH
The Questions Behind The Work.
Building is a way of asking questions. We use short working notes to explain the ideas, examine trade-offs, and connect what we do to existing research.
WORKING NOTES
Ideas worth testing.
Short working notes that make our research questions, methods, and relevant literature easy to examine.
002 / EXPERIMENTAL DIRECTION
When should AI agents work together?
Studying when collaboration improves results—and what routing, retries, and human oversight cost.
Explore the research questions ↗003 / SYSTEMS NOTE
Evidence before completion.
Tracing the difference between running a task, reporting success, and independently checking a result.
Read the systems note ↗RESEARCH THEMES
Across software, people, and worlds.
The work brings together AI systems, human–computer interaction, organizational economics, creative computation, and public-interest technology.
Intelligence with continuity
What should an intelligent system remember, and who can correct it?
Trustworthy autonomous work
When is an automated result reliable enough for someone to use?
Creative systems for more people
Can reusable tools help more people make games, music, and software?
RELATED PUBLICATIONS
Research we build upon.
These are external published papers, not NOAERTH publications. They help frame the questions we want to investigate.
AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation
Multi-agent orchestration and human interaction
Wu et al. · 2024 ↗
Multi-agent orchestration and human interaction
FrugalGPT: How to Use Large Language Models While Reducing Cost and Improving Performance
Efficient selection and routing of models
Chen et al. · 2023 ↗
Efficient selection and routing of models
SWE-bench: Can Language Models Resolve Real-world GitHub Issues?
Independent evaluation of software-engineering tasks
Jimenez et al. · 2024 ↗
Independent evaluation of software-engineering tasks
Generative Agents: Interactive Simulacra of Human Behavior
Memory, behavior, and interactive simulation
Park et al. · 2023 ↗
Memory, behavior, and interactive simulation