Ivan Zografski retweeted
Need is a good reason to reach for AI. Habit is a slippery one. Read that twice.
🔗 solveo.co/post/use-ai-becaus…
Ivan Zografski retweeted
one engineer built a private API endpoint to figure out what his cloud code reviews cost. now it’s one prompt to an agent. blog.kilo.ai/p/mcp
Ivan Zografski retweeted
Writing code is only one part of software engineering. Understanding the problem, making trade-offs, debugging production, and knowing what shouldn't be built are part of the job too. LLMs keep getting better at writing code. That doesn't make engineering judgment obsolete.
before we build any agent for a client, it has to pass 3 questions.
most ideas die at question 2. that's the point.
the filter:
1. who does this today, and how many hours a week does it eat?
2. what number moves if the agent works?
3. who signs off on what the agent produces?
no hours, no number, no owner: no agent.
Ivan Zografski retweeted
We’ve just made public 125,091 DICOM radiology images from across my cancer journey (link below). The data is both longitudinal and multi-modal, dated and linked to my treatment timeline, so you can compare imaging across initial diagnosis, standard of care, first remission, “founder mode cancer care”, and second remission. It’s viewable in OHIF, the standard open-source viewer used in radiology research.
Given the long history of applying deep learning to radiology, we’re hoping to see a variety of methods and models applied to my data. If you find something interesting, please reach out to [email protected], or if you know a researcher who should see this, tag them below.
Thanks to Terra LaFerriere for organizing these images, and Kamil Slowikowski for building osteosarc.com into the resource it is.
Ivan Zografski retweeted
No exaggeration, this guy might be the best interview in pro sports right now
worth reading the plan docs before switching: legacy plans keep their included model pool, the new tiers above $8 have no third-party allowance. if your team runs claude or gpt daily, price that at api list first.
it's six pages with no decision in the first line. ai made long memos free to produce, so the only memo worth reading leads with the ask and the number.
Ivan Zografski retweeted
Your dashboard shows latency went up, but it can't tell you which step got slower or why.
You can ask the TraceRoot agent in plain English, something like "look at the traces from the past three hours and suggest code changes to reduce latency." It reads those traces next to your code at the commit that ran them.
Then it tells you which step is slow and what to change, along with the trace IDs and file paths it checked.
Full walkthrough on a Vercel AI SDK example 👇
traceroot.ai/blog/2026-10-01…
Ivan Zografski retweeted
Go for kilocode its faster or more lightweight for local execution plus i can bring my own key, so i am using gpt models with my chatgpt subscription and OpenCode go models , my workflow doesn't stick to one provider
Ivan Zografski retweeted
We went through 1,085 Reddit comments on debugging agents:
👉 313 comments discussed tracing and instrumentation;
👉 295 discussed evals and regression testing.
The frustration lived in the gap between them.
→ finding the one bad run among thousands that look fine
→ tracing it back to the step that caused it, often upstream of where the error showed up
Here's the full research 👇
traceroot.ai/blog/2026-10-06…
