CNET Test Finds Kudos Bill Agent Stalls at Verification
CNET tested Kudos's AI bill-negotiation agent on AT&T and Verizon accounts and found that it identified possible monthly savings but could not apply them. One-time-passcode and PIN handling interrupted the calls, while the agent's requirement for explicit permission prevented it from accepting plan and autopay changes. Kudos says the Premium feature supports about 60 phone, internet, cable and related providers.
CNET tested the Kudos AI bill-negotiation agent on AT&T and Verizon accounts and found that it could reach provider representatives and identify possible savings, but verification and permission controls prevented it from completing either change. The August 4 review concluded that the tool was not faster or easier than calling customer service for these two attempts.
Kudos markets the feature to Premium members and says its agent can negotiate phone, internet and cable bills. CNET reported that the service connects with about 60 providers and is intended for companies that have retention departments and discretionary offers.
What happened in the two calls
For the AT&T account, the agent reached a representative who said no discount was available on the existing plan but offered a different plan with the same features at $50 per line, compared with the current $71 per line. Kudos declined the change because the user had not explicitly authorized a plan switch. A later attempt failed when the account passcode was entered in the wrong verification field, according to support information reported by CNET.
The Verizon attempt produced an offer for a $10 monthly discount plus another $10 for enabling autopay. The agent would not accept the autopay change without permission, and it could not apply the remaining discount after again running into verification. Kudos support told CNET the potential saving would have been $20 per month, or $240 per year, if the passcode and custom instructions had worked.
Those figures describe offers found during one reporter's tests, not a general savings rate. No bill reduction was completed in either case.
The execution boundary matters
The test separates conversational ability from task completion. The agent could navigate provider calls and surface offers, but the workflow depended on time-sensitive identity checks and user authorization for changes such as switching plans or enabling autopay. Those controls are designed to protect the account holder, yet they also limit how autonomous a consumer agent can be.
Kudos's product page lists supported mobile, home-internet, television and home-security providers and says users keep the savings negotiated by the agent. That page documents the product's intended scope; CNET's hands-on test supplies the evidence about how the workflow performed in these two cases.
For teams evaluating task-oriented agents, success metrics should extend beyond whether an agent completes a conversation. Authentication handoffs, permission capture, recovery from expired credentials and escalation to a human can decide whether the promised outcome is actually delivered.
Key Points
- 1CNET's two tests found possible AT&T and Verizon savings, but the Kudos agent did not complete either account change.
- 2One-time-passcode and PIN handling interrupted the workflow, while explicit-permission rules blocked plan and autopay changes.
- 3The test shows that authentication and authorization handoffs, not only voice quality, determine whether a consumer agent completes its task.
Scoring Rationale
The hands-on test documents a practical failure mode for consumer agents: the system surfaced savings but could not cross authentication and authorization boundaries. The finding is useful for agent design and evaluation, though it comes from two account tests of a narrow consumer product.
Sources
Public references used for this report.
Practice with real Telecom & ISP data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Telecom & ISP problems
