It sounds like you're dealing with the common challenge of ensuring quality for complex voice agents, which can be very time-consuming with manual testing. Instead of hiring someone for hours of daily manual calls, you might consider automating your conversation tests. This allows you to script various caller personas and scenarios, run them against your agent on every deploy, and quickly catch regressions in accuracy or task completion. If you're looking to automate this process, platforms like cekura.ai are built specifically for end-to-end voice AI agent testing and monitoring.