← Back
Single agents
A single agent is one model with one set of instructions, working alone. These are the two things you can change and measure. What you set here is exactly what runs live.
Model
Which AI model runs this agent. Swap it to test whether a cheaper or stronger model changes reliability and cost.
Instructions
Exactly what the agent is told to do — this is the system prompt that runs live. Reword it to test how much the instructions affect the result.
Data connectorscoming soon
Connect this agent to real databases, APIs, and tools, so the same tests run against live data connections.