Using Claude to Build a Bespoke Bluetooth Signal Strength Meter

AI-Driven Bespoke Software for Immediate Problem Solving

Ben Zhang successfully located a lost phone in an office environment by using Claude to develop a custom Bluetooth signal strength meter in approximately one minute. This approach was necessary because the device's Mobile Device Management (MDM) configuration had disabled the standard "Find My" services, rendering traditional tracking methods ineffective.

Technical Implementation: Bluetooth RSSI Tracking

To find the device, Zhang utilized the concept of Received Signal Strength Indication (RSSI). By tracking the strength of the Bluetooth signal, a user can determine their proximity to a target device; as the user moves closer to the device, the signal strength value increases.

Zhang shared the resulting code on GitHub at github.com/ben-z/findphone, demonstrating how LLMs can now be used to generate functional, single-purpose tools for immediate physical-world problems.

The Shift Toward "Bespoke Personal Software"

This incident has sparked a broader discussion regarding the transition from searching for existing software to generating custom tools on demand.

The "Bespoke" Software Trend

Several users highlighted a shift toward creating highly specific, one-time-use applications:

• Rapid Prototyping: Users reported creating games for children, custom video compositors, and e-commerce landing page builders using AI, noting that the "last 20%" of development—the hardest part—is now significantly easier.
• Disposable Tools: Some developers now maintain GitHub repositories filled with "little things" generated by AI that are used once and then abandoned, treating code as a disposable utility rather than a long-term asset.
• Complex Debugging: AI is being used to analyze stack traces and identify bugs in software (such as GIMP) without requiring the user to have the full source repository locally.

Counterpoints and Technical Critiques

While the speed of development is praised, technical critics and experienced engineers raised several points:

• Redundancy: Some noted that Bluetooth RSSI meters are common features in existing hardware (e.g., Garmin watches) and available as numerous Android apps, suggesting the AI was "reinventing the wheel."
• Obvious Alternatives: Critics pointed out that simpler solutions, such as calling the phone or checking WiFi signal strength via an access point (AP) management console, would have been more efficient than writing code.
• Code Quality: There are concerns that the rapid generation of AI code leads to "spaghetti mess" architectures, which may create a significant technical debt burden in the future.

Comparative Use Cases for Signal Tracking

Beyond finding lost phones, the community shared various applications of signal strength tracking:

Method	Use Case	Implementation
Bluetooth RSSI	Finding lost earbuds or Fitbits in brush/woods	Custom meters or specialized apps
WiFi RSSI	Locating children's devices in a home	UniFi WiFi AP signal strength indicators
BLE (Bluetooth Low Energy)	Room presence detection	ESP32s and Shelly switches integrated with Home Assistant