Free newsletter · weekly
The Top K
The k things in AI worth your time each week: news, deep dives and hands-on tests. Real numbers, no hype.
by Sebastian Buzdugan · 21,000+ followers across YouTube, X and MediumFree. One email a week. Unsubscribe in one click.
What you get each week
k = 1
The news that matters
The launches and releases that change how you build, and why.
k = 2
One deep dive
A paper, model or system explained properly, without the hype.
k = 3
One hands-on test
I run the agent, model or tool myself and show the real numbers.
All issues
Oct 22, 2026 · News
Claude Duet Is an Agent That Messages You First
Anthropic just shipped an agent that does not wait for your prompt. It reads your inbox every few hours, drafts the work, and pings you only when it needs a yes.
Oct 1, 2026 · Deep Dive
Top 5 Enterprise AI Gateways (2026): Bifrost vs Kong vs LiteLLM vs Cloudflare vs Vercel
Five enterprise AI gateways compared on deployment, licensing, governance, failover and cost, plus a hands-on overhead and failover test.
Sep 20, 2026 · Deep Dive
Two Models Talking Through KV Cache Instead of Text
An ICLR 2026 paper skips text between models. The most interesting number is in the ablation table.
Jul 31, 2026 · Hands-on Test
Hivemind Wrote 19 Skills From My Coding History. The Best Part Was What It Refused to Write.
A hands-on review of Activeloop's Hivemind: 19 skills mined from my real Claude Code history, the pattern it refused to write, and the recall bug I traced and fixed.
Jul 13, 2026 · Deep Dive
Your Agent Benchmark Is Lying to You
The hard part of agent evals is the sandbox.
Jul 1, 2026 · Deep Dive
Snapshot Once, Rollout a Thousand Times
Your GPUs aren't the RL bottleneck. Rebuilding the environment is.
Oct 16, 2025 · Deep Dive
AI Crawlers Are the New GPUs: Why Data Quality Beats Compute Power
Everyone talks about fine-tuning and GPUs, but the next big leap in AI might start with better crawlers, not faster chips.
