*Hey, Luca here! Welcome to a new edition of the *
This newsletter is brought to you by Unblocked!
AI is firmly in your engineering workflow. But while the token spend shows it, the throughput doesn’t. Humans are very much still in the loop, and that’s likely a context problem.
Our friends at Unblocked will hold a ** free webinar on Aug 19** to cover:
How to track
AI productivityHow
context maturityaffects where teams stallWhat it takes to get
real valuefrom your agents
They know what they are doing and I personally use Unblocked everyday for Tolaria! Recommended.
When a product grows, feedback starts arriving from everywhere.
In my case with Tolaria, it comes from GitHub issues, support emails, community threads, crash reports, the Canny board, random DMs, and of course, my own ideas.
It is tempting to put all of this into one giant backlog, but bugs and feature requests create very different kinds of work.
A
bugmostly needs reproduction, prioritization, and a fix. If it is real, there is usually not much product strategy to debate. The job is to understand what is broken and make it not broken anymore.A
feature requestis different. It needs deduplication, discussion, roadmap judgment, planned / in-progress states, and possibly a place where users can follow what happens next. It also benefits from users voting on actual proposals, so you can use that for broad prioritization.
If you collapse these channels into one, like many teams I know actually do, the system can get noisy very quickly, like:
Users filing feature requests as bugs
Bugs appearing in product boards
Duplicates multiplying everywhere
Engineers spending time routing work manually
About all of this, what works for me is explicit traffic control. Route people to the right place when they submit feedback, and then run an intake workflow that moves misplaced items, closes duplicates, and turns valid work into tasks 👇
I wrote more about this in the full piece on how I run Tolaria as a big open source project 👇
Behavioral interviews can feel vague because there is no clear and immediate feedback. You say something about your career, someone nods, asks for a follow-up, and you leave the call wondering whether the answer was any good.
In my conversation with Austen McDonald, former hiring committee chair at Meta, he gave practical advice to reduce this ambiguity: most behavioral interviews eventually orbit around three questions.
🙋♂️
Tell me about yourself🗃️
Tell me about a project⚔️
Tell me about a conflict
This does not mean you should memorize three canned answers and recite them everywhere. The point is to understand your own career and adapt the same strong stories to many different questions.
For “tell me about yourself”, Austen suggests three parts: who you are, your greatest hits, and why this specific role makes sense as the next step. The goal is that, if the call drops after your intro, the interviewer will already want to talk again.
For a project story, you should guide the interviewer toward the best material before time runs out. So you should signpost the important moments upfront (e.g. the architecture choice, the difficult implementation, the launch tradeoff) and then go deeper where the signal is strongest.
For conflict, Austen says the most common mistake is skipping the actual resolution. Interviewers do not only want to hear that there was disagreement. They want to understand what you did: did you go directly to the person? Did you use Slack, a meeting, a doc? Did you show empathy for the other side? What changed afterwards?
Good behavioral prep is about taking your real experience and making it legible for your interviewer.
You can find the full conversation with Austen here 👇
Finally, here are the best articles I have read this week:
A very useful counterweight to the usual AI productivity discourse. Annie shows that engineers can feel more productive while their actual experience gets worse, especially around flow state. An interesting second-order effect leaders should watch before declaring victory.
Short and sharp argument against one of the laziest takes of the AI era: that we do not need juniors anymore. Frederick acknowledges the “succession” point, but counterbalances it with the innovation one: juniors bring fresh takes on many things, and stress-test your assumptions because they have not yet learned all your bad habits.
Tiny Rands post about why writing matters: because writing is a way of building. I love this framing. In a world where we can generate more words and more code than ever, the most valuable part is still the thinking you force yourself to do before you publish or ship.
And that’s it for today! If you are finding this newsletter valuable, subscribe to the full version!
1700+ engineers and managers have joined already, and they receive our flagship weekly long-form articles about how to ship faster and work better together! Learn more about the benefits of the paid plan here.
See you next week!
Luca
