RESEARCH

Research

Ideas before products. We publish research on AI systems infrastructure, decision models, routing, security, and agent execution.

RESEARCH / 006

Workflow Drift

Measuring behavioural changes in agent systems when the underlying model changes. October 2026.

READ PAPER →

RESEARCH / 005

Efficient Inference at the Edge

How much intelligence can move toward local compute? Benchmarks across consumer hardware.

READ PAPER →

RESEARCH / 004

In-Band AI Security

Can security participate directly in agent execution? A framework for runtime safety in autonomous systems.

READ PAPER →

RESEARCH / 003

Agent Isolation

How should autonomous workloads execute safely? A survey of isolation techniques for AI agents.

READ PAPER →

RESEARCH / 002

Decision Models

Can small encoders replace generative models for bounded machine decisions? Benchmarks and architecture.

READ PAPER →

RESEARCH / 001

Model Routing

What happens when the optimal model changes? A study of routing dynamics in multi-model systems.

READ PAPER →