This episode of DataFramed explores how AI is transforming software telemetry—the systems that track application performance and user behavior—and how telemetry is critical for managing AI-driven software development. Guests Ledion Bitincka and Nikhil Mungel from Cribl discuss the challenges of exponential data growth, the shift from human workers to agents, and the importance of focus and judgment in the age of AI-generated code.
Summarized by Podsumo
Software telemetry is becoming crucial as AI exponentially increases the volume and complexity of data generated by applications, including conversational and agentic traces.
The role of engineers is shifting from 'knowledge workers' to 'judgment workers' who specify problems and verify AI-generated solutions rather than writing code from scratch.
AI can amplify both good and bad habits; staying focused on shipping value rather than starting multiple projects is essential to avoid waste and high token costs.
Cribl's custom app builder empowers subject matter experts to create tailored workflows on top of telemetry data without waiting for engineering teams.
Cutting-edge research is underway to process telemetry data using GPU-based models, which could dramatically reduce computational costs.
"Software telemetry is data and information that any software system produces to log actions, aggregate metrics, or trace distributed actions. It's the audit trail your software leaves for every decision it takes in production."
"Humans are moving away from being knowledge workers to judgment workers. Your humanness and attention to detail in understanding what the business needs is still very much dependent on a human being."
"AI is amplifying both the good and the bad. If you're somebody that drills deep and has a tendency to go into the weeds, AI is there to help you dig that hole even deeper."
"Shipping value is key. I don't care if we made progress on 10 things during the month; if we haven't shipped something to the customer, we haven't shipped value."