'I closed my inbox and waited to see how the afternoon went' - I gave ChatGPT permission to watch my inbox, then stopped checking my email

'I closed my inbox and waited to see how the afternoon went' - I gave ChatGPT permission to watch my inbox, then stopped checking my email Event-triggered tasks promise a calmer way to handle email ChatGPT Work’s new event-triggered scheduled tasks offer an intriguing alternative to the compulsive email checking that fills my days. The feature essentially acts as an AI-powered secretary, screening your Gmail inbox, Slack channels, and other communication channels, then alerting you or responding as you request in a prompt. I decided to use the feature as an aggressively selective email assistant. ChatGPT could watch everything arriving in my inbox, but it was only allowed to interrupt me when a message created an actual obligation, and it should suggest how to respond and have a draft ready to go. Event-triggered tasks are available on eligible plans in ChatGPT on the web and mobile. I used ChatGPT on the web and connected my Gmail account when prompted. You can see the prompt I used below. The current event triggers support new Gmail messages, which can optionally be narrowed by sender or subject. I wanted ChatGPT to judge every incoming message by its contents, so I left the trigger broad and made the instructions selective. That distinction matters. The trigger decides when ChatGPT wakes up, while the prompt decides what it does once awake. In my case, every new email could prompt a run, but only messages meeting my criteria were supposed to produce an alert. I closed my inbox and waited to see how the afternoon went. Here's how to do it My original prompt was fairly basic. I told ChatGPT, "When I receive an email that requires me to do something, or comes from a family member, tell me what I need to do and draft a potential reply." Sign up for breaking news, reviews, opinion, top tech deals, and more. It still left plenty of room for interpretation, and I was wary of spam getting listed as something I needed to do something about, but it came down to being a good test for ChatGPT. After a bit, I did clarify to ChatGPT that it should alert me about any email with a single sentence explaining who was writing and what they might need from me. I also made sure to add that the AI shouldn't reply without my input. I also opened ChatGPT’s notification settings and made sure task alerts could reach me outside the inbox. Depending on the account and notification category, ChatGPT can offer push, email, or text message delivery. I avoided email for obvious reasons, since an email announcing that ChatGPT had found an email would have possibly confused the AI. With the task running, I shut Gmail and returned to everything else I had planned for the afternoon. For the first several minutes, I still felt the pull to check, but eventually, I stopped thinking about it. The absence of constant email checks made the afternoon feel longer in a good way. I could finish a task without repeatedly glancing at another tab to make sure the internet had not developed an emergency specifically for me. Only 3 interruptions The first ChatGPT alert concerned a message from my bank about two hours after setting up the alert. ChatGPT identified the sender, explained why the email appeared to require attention, and told me what I needed to review. I opened that message directly, made sure it didn't actually require a response, and closed down the inbox. The second alert came from a family member. This was a more interesting test because family messages rarely contain the tidy language that makes conventional email rules useful. They can require a decision or a quick response without the more business-style setup and keywords in professional messages. Still, ChatGPT recognized that the message expected something from me and brought it to my attention and wrote a brief reply draft for me to approve. I added a little, but ChatGPT nailed the main points I needed to make, and it wasn't the kind of email that needed a lot of personal detail, so a few minutes after the alert arrived, the response was out the door. The last alert concerned tickets I had recently purchased and a change to the itinerary for an upcoming event. In this case, the alert was very necessary as it changed some of the other plans for that day, and I had to let others know of the shift. That wasn't in ChatGPT's suggested response, of course, but just getting the heads up was useful. Those were the only three times I looked at Gmail all afternoon. Other messages arrived, but ChatGPT did not summon me to look at them. When I went to my inbox afterward, I clocked only two emails that I might have wanted to deal with earlier, and neither had to be handled immediately. Nothing terrible happened because I didn't see a promotional email at the moment of delivery. Anyone trying this should begin with a short test period and review the first few runs in ChatGPT’s Scheduled section. If ordinary receipts keep slipping through, add a clearer exclusion. If requests from real people are being missed, explain what counts as an expected response and allow ChatGPT to flag uncertain cases. I would also keep the first version informational. Ask ChatGPT to summarize the obligation and point you toward the email, then make the decision yourself. It's a feature worth trialing. Many AI tools try to prove their value by producing more text, more summaries, and more notifications. This one worked because it knew when to leave me alone. Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. ➡️ Read our full guide to the best business laptops 1. Best overall: Dell 14 Premium 2. Best on a budget: Acer Aspire 5 3. Best MacBook: Apple MacBook Pro 14-inch (M4) Eric Hal Schwartz is a freelance writer for TechRadar with more than 15 years of experience covering the intersection of the world and technology. For the last five years, he served as head writer for Voicebot.ai and was on the leading edge of reporting on generative AI and large language models. He's since become an expert on the products of generative AI models, such as OpenAI’s ChatGPT, Anthropic’s Claude, Google Gemini, and every other synthetic media tool. His experience runs the gamut of media, including print, digital, broadcast, and live events. Now, he's continuing to tell the stories people want and need to hear about the rapidly evolving AI space and its impact on their lives. Eric is based in New York City. You must confirm your public display name before commenting Please logout and then login again, you will then be prompted to enter your display name.