Sandeep B.
Core Paid Feature Never Worked, Support Failed to Resolve It, and Refund Was Denied
I strongly recommend that prospective customers think twice before purchasing an annual subscription from Final Round AI. I bought the annual plan primarily for the spoken-answer analysis feature because I had used it successfully on another account. On this paid account, the feature never worked. The failure was consistent: speech-to-text worked, but clicking “Finish” caused the analysis to fail. I went far beyond what any paying customer should have to do and identified the backend issue myself: the /api/mock-interview endpoint was returning a 500 Internal Server Error. I provided screenshots, timestamps, reproduction steps, the failing endpoint, and the exact error response. Instead of real technical support, I received a stream of repetitive, generic, AI-style replies asking whether the issue was resolved or requesting information I had already provided. There was no clear explanation of what had actually been investigated, fixed, deployed, or tested. My original request was not for a refund. I simply wanted the feature I paid for to work. I requested a refund only after the issue remained unresolved and support failed to provide any meaningful progress. The refund was then denied using a rigid 3-day policy and session-count rules, even though support admitted they could not determine whether those sessions were successful use or repeated attempts to troubleshoot the broken feature. They also refused to escalate the matter to a human billing manager and treated the decision as final. So the outcome was simple: the core feature never worked, the issue was not fixed, meaningful human support was unavailable, and the refund was denied. In my experience, this felt like being trapped in an annual subscription while dealing with automated responses instead of accountable customer support. My advice: do not commit to an annual plan unless you are completely certain the product works for your use case. There are other interview-preparation tools available, and I would strongly recommend choosing one with transparent support, real human escalation, and a fair approach when a core paid feature fails.



