The Mistakes You Can’t Take Back
The last one is the most hard hitting - The decision you make by not deciding. The conclusion is worth reiterating - Whichever doors you walk through, walk through them on purpose.
Bookmarks RSS feed
-
-
Inside Thinking Machines’ Interaction Models
Moving interactivity from harness to model itself is clever, no doubt. What’s more interesting is how they solved constructing responses from partial inputs. And how responses from bigger background model is stitched back in interaction model.
-
How AI Agents Manage Memory and Avoid Forgetfulness
Great deep dive on the deceptively simple concept of llm/agent memory. Biggest takeaway: it’s a game of accurate retrieval rather than storage.
-
An Ex-Meta L8’s Agentic Engineering Setup
It’s always fascinating to see agent setups people use to ship code to production. I tend to err on the side of using OOTB/native tooling unless something really comes in my way. If I were to pick a couple of improvements to my workflow based on the article, it’d be interactive HTML plans and agentic code reviews.
-
When Was the Last Time You Did Just One Thing?
Captures the essence of Slow Productivity but with sharp and memorable anecdotes. I could personally relate to each and every example.
-
https://blog.bytebytego.com/p/token-spend-out-of-control-the-case?triedRedirect=true
Routing is a great method of reducing token spend. Using a society of task-specific small language models is another. Both work to ensure the 80% light repeatable tasks don’t wastefully end up with fat models and only the genuine 20% do. Different techniques, different dynamics, different tradeoffs.
-
Your Job Is Disappearing. I Watched Mine Disappear At Least Six Times.
My favorite quote - “Tasks run out. If you are in a problem-solving role, you’ll know that you never run out of problems. The type of problems you solve today may fade away, but the supply of problems will not.”
-
How DoorDash Built a Testing System to Evaluate LLMs
Not as path breaking as the intro makes it sound. It’s just prompt engineering and evals at scale, for godsake.
-
AI’s impact on software engineers in 2026: key trends, Part 2
AI tooling addiction is real, guys. Please consider the productivity and quality tradeoffs and use AI responsibly :)
-
How Grab is Using AI Agents to Boost Team Productivity
Most software teams are building wrappers around LLMs to automate their processes. That creates suboptimal solutions more often than not. Grab’s data engineering case study is a great example of crafting a well-engineered system of agents to optimise an internal bottleneck that uses LLM as a component rather than its brain.