This Week's Signal: The skill that closes the AI training gap

Heads up: this one's an ad.

Power users figured this out early: speaking a prompt gives you 10x more context in half the time. You include the edge cases, the examples, the tone you want — because talking is fast enough that you don't skip them.

Wispr Flow captures everything you say and turns it into clean, structured text for any AI tool. Speak messy. Get polished input. Paste into ChatGPT, Claude, Cursor, or wherever you work.

89% of messages sent with zero edits. 4x faster than typing. Works system-wide on Mac, Windows, and iPhone.

Start flowing free

Last week I named the AI training gap: organizations buying tools faster than their teams learn to use them, with results landing all over the place. I said to close it starting with yourself. A few of you wrote back with the obvious next question: which skill, exactly?

Here's my answer. If I could teach a Chief of Staff one AI skill, it would be context engineering. It's the difference between output you have to fix and output you can stand behind. And this week a guest contributor built a reusable tool that does it for you.

A founder I work with asked me for a technical scope and business case for a project the rest of the leadership team hadn't approved yet. I gave AI the ask straight: "draft a technical scope and business case for this project."

What came back was confident and off. It treated the project like a done deal, baked in commitments leadership never agreed to, and used a tone that would have landed badly in that room.

So I gave the model the frame first: this is still pending approval; here's what's actually decided versus what's open; here's who’s skeptical and why; and here's the line I can't cross in committing resources.

Same request. The second draft was something I could hand the founder to walk into the room with.

When the output is wrong, look at the context you skipped before the words you chose.

Want more?

This week's skill comes from Georgia Hirth, who taught herself to build no-code AI agents and built a reusable tool for exactly this.

Most people have heard of prompt engineering. Context engineering is the part that fixes bad output. It's the practice of deciding what goes into the AI's context window before it starts reasoning: your role definition, constraints, required format, and the information you deliberately leave out. Prompt engineering asks how to phrase the request. Context engineering asks what the model needs to know to do the job well. Leave those gaps open, and the model fills them with defaults that have nothing to do with your situation.

For a CoS, that matters more than for almost any role. The ask is never one thing. It's "hold this context, know these people, produce something I can stand behind." Without the right frame, the model guesses.

Georgia built a Claude skill that does the framing for you. You paste in a rough task; it audits the gaps and then returns a tight context layer with role clarity, constraints, format, and a definition of done. Set it up once in Claude, and it's there every time. For ChatGPT, Copilot, or Gemini, it works as a one-off instruction.

Here's the context layer it produces from a rough briefing task. Copy it, fill the brackets, and paste it in before your real request:

Use it whenever the task is high-stakes and the model needs more than a single sentence can carry.

Georgia made the full skill freely available on GitHub.

You can follow Georgia on LinkedIn here.

Gartner now frames context engineering as the discipline that replaces prompt engineering for enterprise AI: structuring the data and environment so that AI understands intent, rather than relying on manual prompts. For a CoS, this is familiar ground. The value is in how well you frame the problem, which is exactly the judgment this role already runs on.

Anthropic's guide, Effective Context Engineering for AI Agents, is the deeper primer if you want to go past this issue. It's free, and it's the read that makes the template above stick.

If you know another CoS sitting in the "we have the tools, now what?" conversation, forward this to them.

Heads up: this one's an ad.

You don't have to scroll every AI thread, track every new tool, or watch every demo. 

The Rundown AI breaks it all down for you — the latest AI news, tools, and tutorials in one free 5-minute email every morning. 

Trusted by 2M+ professionals at Apple, Google, and NASA.

Sign up to start learning.