Barge Test: interruption and disclosure testing for AI debt-collection calls

Both halves of this idea are already taken. For the adversarial caller suite, horizontal voice testing platforms (Coval, Hamming, Cekura, Bluejay, Future AGI) already simulate callers who barge in, get angry and set compliance traps, and Coval is well funded. For the recorded-call scorer, collections QA incumbents (Prodigal, Sedric, CallCabinet) already score every call for Reg F and FDCPA issues, so that product is close to a commodity. Collections voice vendors such as Salient, Skit and Dograh also claim they do compliance testing and monitoring themselves. I found no independent, collections-specific adversarial test suite. The remaining opening would be a pre-built, maintained library of debtor scenarios tied to Reg F and state rules: cease requests made mid-sentence, a dispute raised while the agent is still delivering the mini-Miranda, requests to call at a different time or place, attorney representation, NYC SHIELD's 3-in-7 call limit versus the federal 7-in-7, and third-party contact. That library would come with a scored report a compliance officer could show auditors, clients or the CFPB. The likely buyers are agencies picking among several voice AI vendors and the creditors who audit those agencies. They would want a neutral party certifying any vendor, which is something neither the horizontal testing tools nor the vendors testing themselves can claim. The risks are that Coval or Hamming ship a collections template pack, and that the market is small and narrow. So this probably works best as a narrow wedge or a certification-style service, not a standalone platform, and the recorded-call scorer should be dropped or folded in.

• Coval

  Well-funded general voice agent simulation platform (it reports a $28M Series A). It runs persona-based simulated callers with barge-in, irate callers and compliance traps. Not specific to collections, but it can run the adversarial caller suite today if the customer writes the scenarios.

• Hamming

  Automated QA and production monitoring for voice agents, with SOC 2 Type II and HIPAA and a focus on regulated enterprises. Horizontal, with no collections rule pack found.

• Cekura

  Voice agent testing that covers interruptions, latency, IVR navigation and voicemail, with a self-serve developer tier. Horizontal.

• Bluejay

  YC X25 testing and observability platform for voice and chat agents. Horizontal.

• Future AGI Simulate

  Simulated real-call testing for voice agents. Horizontal.

• Prodigal (ProInsight)

  Consumer-finance conversation intelligence that scores every call for Reg F and FDCPA compliance and now also sells AI voice agents. This directly overlaps the recorded-call scorer for lenders and agencies.

• Sedric

  Compliance monitoring for financial services calls. Its rule library includes dispute handling, so it overlaps the recorded-call scorer.

• CallCabinet

  Call recording and QA that says it automatically detects compliance violations, including Reg F and TCPA.

• Salient

  AI collections voice agent vendor that claims built-in FDCPA, Reg F and TCPA compliance and real-time compliance monitoring. A likely buyer, but it also tests in-house.

• Skit.ai

  Large AI collections voice platform that lists embedded compliance monitoring and adversarial input guardrails. A potential buyer, or it may self-serve.

• Dograh

  Collections voice agent vendor that markets an AI-to-AI framework simulating debtor scenarios for compliance readiness before deployment. This is the closest thing to a collections-specific simulated-debtor test, but it only covers its own agents.

Checked 2026-10-17