AgentArch
Fits teams choosing an agent architecture for enterprise work. Borrow the grid of 18 configurations (orchestration, memory, thinking, tool style) as an ablation plan.
What it asks the agent to do
- complete a simple enterprise request with tools
- complete a complex multi-step enterprise workflow
- compare orchestration, memory and reasoning settings on the same task
What to borrow from it
taxonomy
Caveats
Only two tasks; no public code; results are about architectures, not a reusable test set.
Details
| Workloads | workflow automation, tool calling, it ops |
|---|---|
| Interaction | single turn |
| Environment | api sandbox |
| End state checked | No |
| Tools | enterprise tools in a simulated setting |
| Grading | programmatic check |
| Size | 2 |
| Data | synthetic |
| Personal data | none |
| License | unknown |
| Harness | none |
| Status | active |