SIAgentNews.com
← All news

How a user used Claude to coach them through a live internet bill negotiation

A user successfully cut their monthly internet bill in half by using AI to research, script, and provide real-time coaching during a retention call.

· First reported by patrick-reagan.com

Claymation: picture for the story 'How a user used Claude to coach them through a live internet bill nego'
AI-made claymation for this story. No real people or products are shown.
$600amount saved per year

The short version

  • The user saved $600 annually by reducing their internet bill by 50%.
  • Claude was used to research competitor offers, customer forums, and retention tactics.
  • The user built a custom app using Twilio to handle live transcription of the call.
  • The AI provided real-time coaching via a terminal while the user spoke to a representative.
  • The representative agreed to the requested price just 75 seconds after the initial pitch.

Why it matters

This demonstrates how LLMs can move beyond content creation into real-time personal advocacy. It shows that AI can act as a specialized consultant for complex, high-friction tasks like navigating corporate bureaucracy.

The goal of the negotiation

A user noticed a significant discrepancy in their internet service costs. While they were paying for a top-tier plan, a competitor was offering the same speeds at their address for half the price. They decided to use Claude, an AI tool, to see if they could secure a lower rate without actually switching providers.

The user identified a common problem where long-time customers often pay more than new sign-ups. They wanted to see if AI could help them navigate the specific hurdles of a retention call. The goal was to leverage information about competitor pricing and customer behavior to force a discount.

The project began with a failed attempt at full automation. The user tried to write a small app to dial the provider and navigate the phone menu using keypad presses. However, the provider's system did not recognize the call as coming from a customer, leading the user to pivot toward a more manual, AI-assisted approach.

Researching the leverage

Before making the call, the user collaborated with Claude to build a comprehensive strategy. They started by pulling their current bill and identifying exactly how long they had been a customer. They also gathered data on what competitors were offering at their specific location.

To understand the human element of the negotiation, the user had Claude analyze customer forums and blog posts. They looked for specific tactics that other people had used successfully to get lower rates. This helped the AI identify common reasons why customers were turned down, such as asking for specific new-customer promotions by name.

Based on this research, Claude wrote a specific pitch for the user to lead with. They also developed a list of likely counteroffers from the provider. This prepared the user to respond to common objections immediately without having to think on their feet during the high-pressure conversation.

Live transcription and coaching

The user decided that a static script might not be enough because phone calls rarely follow a linear path. They wanted the AI to hear what the representative was saying in real-time. To achieve this, they set up a system using a Twilio number and a live transcription service.

The user dialed the provider from their own cell phone through the Twilio number. This allowed the provider's system to recognize them as a registered customer, bypassing the menu issues they faced earlier. The app then transcribed both sides of the conversation as it happened.

This transcript was fed into a Claude Code session running in a separate terminal. As the representative spoke, the AI analyzed the text and provided coaching instructions to the user. This setup allowed the user to stay on the line while receiving live guidance on how to handle the representative's specific responses.

The results of the call

The actual negotiation took place on October 26th and lasted just over ten minutes. Because the user had a clear pitch and a prepared list of counters, the conversation moved quickly. The representative offered the exact price the user requested only 75 seconds after the initial pitch was made.

The user was able to accept the new price immediately via an emailed link before hanging up. The system worked as intended, moving the user from a high-tier price to a much more favorable rate. The user confirmed the success when the next bill estimate on October 28th reflected the new, lower price.

This successful interaction highlights the utility of using AI as a real-time coach. Instead of just generating a one-time script, the user created a dynamic feedback loop. This allowed them to handle an unpredictable human interaction with the confidence of having a researcher and strategist by their side.

What this means for users

This story illustrates a shift in how people can use AI for 'everyday chores.' While many use these tools for writing or coding, they are increasingly capable of handling complex administrative tasks. By combining AI with transcription tools, users can bridge the gap between digital intelligence and real-world phone interactions.

For the average consumer, this means that navigating corporate bureaucracy might become easier. Tasks like appealing property taxes, planning for college, or lowering utility bills can be broken down into research and real-time coaching. It turns a stressful social interaction into a structured, data-driven process.

As these tools become more accessible, we can expect to see more 'lo-fi' versions of customer service automation. Users who are not engineers can still build functional systems by connecting different APIs. This democratizes the ability to use technology to save money and time on routine life management.

Sources and links

First reported by patrick-reagan.com. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

More AI news