All work
LOG-05In build2026
AI-to-AI call lab: what happens when a customer's agent calls your voicebot
A test bench where a customer-side AI agent calls an enterprise voicebot. Measuring loops, mutual detection, fraud surface and cost.
- Context
- Research prototype · follows up my paper on AI-to-AI calls
- Role
- Research question, test harness, analysis
- Shows
- Thinking one step ahead of the customer: the next failure mode of voice deployments.
The question
Customers are starting to delegate calls to their own AI assistants. Enterprise voicebots were designed for humans on the other end. What happens when both sides are machines?
What the lab measures
- Loops. Two agents politely asking each other to repeat, forever, on a metered line.
- Mutual detection. Can either side tell it is talking to an AI, and should it change behaviour when it does?
- Fraud surface. Authentication flows built for humans, probed by an agent that never gets tired.
- Cost. What AI-to-AI traffic does to per-minute pricing and to the business case.
This entry will be updated as experiments run.