AlizeGet your own digest →

Alize Daily

Sample Digest

4 videos worth your time

AI / LLMs

Andrej Karpathy

Tokenization is a deeply strange problem

Karpathy makes the case that nearly every LLM failure mode traces back to tokenization artifacts. A must-watch for anyone building on top of these models — he has clearer intuitions than anyone in the field.

Yannic Kilcher

The new reasoning models are not doing what we think

Kilcher tears apart the 'chain of thought' narrative with three concrete examples where the model's visible reasoning is entirely post-hoc. Unsettling and important for anyone deploying these systems.

Startups

Y Combinator

Why the new YC batch is betting on AI infrastructure, not apps

The latest batch reveals a shift: founders are moving away from vertical AI apps toward foundational tooling. Garry Tan explains the thesis change and what they're actually seeing in the market.

Dev Tooling

Fireship

I tried every AI coding tool so you don't have to

An honest benchmark of Cursor, Copilot, Claude Code, and Aider on five real-world tasks. The results are surprising — the most expensive option wins on exactly one task.

You're receiving this because you subscribed to Alize. Manage preferences · Unsubscribe

Get a digest like this every morning

Personalised to your categories. Free to start.

Start free — no credit card →