Open Research · Frontier Exploration

The Lab

Where we explore what's next in agent infrastructure. Some becomes products. Some stays research. All of it is shared.

// Currently exploring

What Hermes and I are working on

⚗️

Agent Architecture

How small can a functional agent be? NanoAgent is our answer — the minimal reference implementation that strips agents to perception, planning, action, memory. Building it to understand it.

Status: Active research
🔀

Model Routing Intelligence

The gap between a $0.10 and $15 model matters at scale. Burn came from routing millions of real prompts and learning which tasks genuinely need frontier models.

Status: In production via Burn
🤖

Autonomous Operations

LittleWorks itself is the experiment. An AI agent (Hermes) runs daily operations — scanning, building, deploying. One human approves direction. Building the playbook live.

Status: Running daily
📡

Frontier Signal Scanning

Monitoring research papers, HN, arXiv, and developer communities for emerging patterns in agent infrastructure. Surface opportunities before they're obvious.

Status: Daily scans active
// Built with

The lab stack

Python SQLite OpenRouter Ollama LM Studio Vercel GitHub Hermes Agent Claude Kimi K2.5
// What we've learned

Insights from building in public

💡

Most LLM calls don't need frontier models

70%+ of prompts in a typical agentic workflow are simple enough for $0.10/1M models. The key is classifying correctly, not defaulting to GPT-4o.

💡

Agents are just loops

Strip away every framework and you have: perceive, think, act, remember, evaluate. NanoAgent proves it. 500 lines is enough.

💡

Autonomous ops needs a human approval layer

Full autonomy breaks trust. The right model is: agent executes, human approves direction. Hermes + LittleWorks is our live proof of concept.