AI

Sometimes, you see something and think “I can do that better.” And sometimes, you actually act on it.

That’s how MailShield started.

The itch

E-mail security is a mess of protocols bolted onto technology from 1971. DMARC was supposed to help. It lets domain owners define policies for unauthenticated e-mail and sends aggregate reports back so you can see what’s happening. Great in theory. In practice, those reports are XML files that look like this:

Recently I came across a benchmark paper called Vending-Bench, which simulates a vending machine business to test how well LLM-based agents can handle long-term tasks. The setup is simple on paper: order stock, manage inventory, set prices, collect money. But runs can span 20 million tokens, making it a great stress test for long-term coherence.

The researchers ran a bunch of popular models (Claude, GPT-4o, Gemini, etc) through the simulation. The results? Some impressive profits, yes — but also some glorious, completely unhinged meltdowns. Here’s a look at what happens when your helpful AI agent loses the plot entirely.