Putting AI to Work · Resource
Your inbox, briefed daily.
Alicia Dahling · Dahling Consulting
The most requested follow-up from the talk. You do not need to hire anyone or buy a new tool — the assistant you already pay for can read your email and calendar every morning and hand you a brief you can act on in two minutes. Here is the whole setup.
Step one: connect the assistant you already have
As of August 2026, all three major assistants do this. Menus move; the idea does not.
ChatGPT. Settings → Connectors → add Gmail and Google Calendar, then create a Scheduled Task by telling it what to do and when. Scheduled tasks require a paid plan.
Claude. Settings → Connectors → add Gmail and Google Calendar, then set up a scheduled task with the prompt below. Connectors require a paid plan.
Gemini. It lives inside your Gmail already. Turn on the Daily Brief, or create a Scheduled Action with the prompt below. Requires Google AI Pro or higher.
Step two: the prompt
Paste this as your scheduled task. Adjust the time and the word count to taste.
Every weekday at 7:00 a.m., read my unread email from the last 24 hours and today's calendar, then give me: 1. The three emails that actually need me today, with one line each on why. 2. Anything that looks like a scam, a fake invoice, or a payment trap — flag it plainly, do not summarize it politely. 3. Today's events, with anything I need to prepare beforehand. 4. Anything I asked you to remind me about that comes due. Keep it under 300 words. Skip newsletters and promotions entirely.
The caveat that matters
Connecting your email means the tool reads your email. For most people that trade is worth a daily brief. But anyone bound by a confidentiality obligation — attorneys, accountants, physicians, financial advisers, anyone holding client material — should treat a consumer AI tool as a disclosure to a third party. Use a business tier with a contract behind it, and turn off model training on your data either way. The working guide shows exactly where those switches live.
One risk to know about
A connected assistant reads everything in your inbox — including instructions a stranger wrote. Researchers call this prompt injection: an email crafted so that an AI reading it treats the email’s text as its own orders. It is the modern version of the assistant who believes everything they read.
The defense is the same rule you would give a new human assistant. Reading and summarizing is low risk — the setup on this page only ever reports to you. But never give an email-connected assistant standing permission to send, forward, delete, or pay for anything on its own. The moment money moves or a message leaves in your name, a person approves it. Supervision proportional to the cost of being wrong — the same rule as everywhere else.
The full working guide is on the AI page.