Weekly Bites #2: Productivity Tracking, Tool Flexibility, and Junior Dev Impact

This week’s bites:

• Productivity Tracking: Why multimodal LLMs could finally make automatic productivity tracking work
• Tool Flexibility: Why you should keep your tool choices flexible when everything changes so fast
• Junior Dev Challenges: The impact of AI on entry-level developer opportunities and what to do about it

Multimodal LLM for productivity tracking

We all want to be more productive. The usual advice is: track what we do. I’ve tried all kind of tools, from plain text file to fancy apps. It never sticks for long. Manual logging is boring, and I forget.

I think the best solution is logging has to happen by itself. Our laptop or phone could notice apps, windows, and calendar, then log what we are doing minute by minute. It might analyze the screen and rate how focused we are for each task. It should run on the device for the best privacy though. With multimodal LLMs, this should be possible by now.

But screens are only part of our day. To cover the rest, we probably need sensors or cameras, like what in the new AR glasses that Google and Meta are developing. They should be capable of summarizing “walking”, “shopping”, or “cooking” without keeping raw video, and rate the quality of the task too. If AI-powered AR glasses catch on, this gets easier. Maybe a few years, maybe sooner.

Keep tool choices flexible

In the LLM era, flexibility is the way to go when picking tools. Things change very fast. Last month your team liked Cursor, today it may be Claude Code, and next month may well be something else. So keep your options open.

Make choices that are easy to undo. Subscribe month to month instead of annual, unless the discount is big and you are sure you will use it every day. Aim for tools with open exports so your data goes with you if you one day decide to switch.

And you should learn new tools just enough for your work. Don’t go too deep; it may be irrelevant in the next couple of months when better tools come out. At the same time, explore and compare so that you are always up to date with the best options.

AI affect junior devs

Junior developers + AI don’t magically equal a senior dev. No matter how powerful the new AI coding tools are, skill still comes from practice: solving messy problems, debugging bad code, figuring things out the hard way. That’s a necessary part of becoming a senior.

Now AI is taking away a lot of those practice opportunities. Many tasks that used to be given to juniors, like small bug fixes, simple features, and routine testing, now get done faster and cheaper by AI with a senior overseeing it. So some companies are hiring fewer junior developers. That could put a whole new generation of young developers in trouble.

What can a Computer Science undergraduate do? My advice is to try to be one of the best students and get an internship at the best companies. I know that might not apply to everyone, but smart and talented students are still sought after, as companies still highly value their potential.