Hi {{first_name|gorgeous}},

I'm embarrassed to admit how long it took me to learn this 😲

As someone who works with AI every day, it was years before I realized:

You can edit prompts 🤯

Instead of this:

You write a prompt. AI gives you a response, but it's not right. So you type another message: "No, what I meant was..."

You can just do this:

Go back to your original prompt and edit it directly. Add what was missing, revise whatever wasn't landing, and resubmit. AI regenerates from the updated version.

This saves you needless back and forth, and cuts down your usage (if you’re in Claude, this adds up fast).

It works in Claude, ChatGPT, and Gemini.

I never knew I could do this until someone else showed me. So let me be that person for you 🥂

Here’s a quick walkthrough:

xx

Emily

💄 UN Women found that 44% of AI systems demonstrate gender bias, and 1 in 5 AI responses about women are outright sexist or misogynistic. Built by the culture that created the problem, optimized to reflect it back at scale.

💄 Anthropic’s Mythos model was sent into classified US government systems as a security test and found real vulnerabilities. The company that makes your caption tool is also poking holes in government infrastructure.

💄 Claude Tag is now in beta, letting you add @Claude as a member of your Slack workspace. It joins channels, builds memory over time, and takes on tasks without being asked. Anthropic writes 65% of its own product code via Claude Tag. Your smallest hire might be your most productive.

💄 Google is investing $75M in A24 to co-build an AI filmmaking lab with DeepMind, developing AI-generated storyboards alongside the studio’s actual directors and crews. The indie studio behind Everything Everywhere All at Once is now a DeepMind research partner.