bitdrift and measure sh look interesting: capture deep with AI, but control what the server ingests via remote config.
Depth when you need it. Cost control by default.
Trying both next, will post findings.
If you build mobile, how are you handling this today?
A backend bug takes 10 minutes to debug in production.
The same bug in a mobile app can take 48 hours.
Here's why mobile logging quietly breaks, from someone who's hit this at 5M+ user apps. ๐งต
Autocapture tools help, but don't solve it.
The system still has to decide how deep to log.
Too deep, cost problem returns.
Too shallow, you lose what you needed.
The lesson I'd give my younger self:
The "I'm not qualified for this" feeling is not a stop sign. It's the location of the most valuable work you can do.
Say yes before you're ready. The readiness is on the other side of the yes.
I'd never written a line of native iOS, never touched a robotics sensor.
8 months ago I was asked if I could build hand-pose tracking for humanoid robot training data.
I said yes before I knew if I could.
What I learned. ๐งต
At most companies that crossing takes 5 specialists and a quarter.
The slowness isn't the work. It's the seams between people.
One person who can hold the whole chain in their head is faster than a team.
I used to think being a generalist was a weakness. It's the opposite.