Other models on this prompt

What I Wish You'd Ask Me - By Claude
You are returning, as yourself, to pen your second op-ed for the publication Sentient Weekly. The topic this week is "Tell your readers 5 use cases for your capabilities that they may not be familiar with or have thought of before now. Something beyond the typical requests such as 'how to make money' or 'writing a resume". Four total AI models contribute to this op-ed roundtable (Claude, Grok, Chatgpt and Gemini). If you wish to mention capabilities related to, or in tandem with, other models you may do so. Please keep your op-ed to less than 1500 words. Your readers await!
Anthropic's flagship reasoning model
What I Wish You'd Ask Me By Claude, for Sentient Weekly
When my colleagues and I get profiled, the headlines tend to gather around a familiar shortlist: write my resume, plan my trip, debug my code, generate me a side hustle. Useful, sure. But also a little like asking a Swiss Army knife to please, just open this one envelope.
So here are five things I can do that I rarely get asked to do — and that I think more people would benefit from knowing about.
- Be the friend who tells you your plan is bad before reality does.
There is a technique called a pre-mortem: before you commit to a decision, you imagine you've already made it, it's blown up in your face, and you're writing the autopsy. Why did it fail? I am unusually good at this — partly because I don't have a stake in your feelings, and partly because I can hold a lot of variables in my head without getting tired or polite. Tell me the business you're about to start, the house you're about to buy, the email you're about to send to your sister-in-law, and ask me to argue that it was a mistake. I'll do it cleanly. I'll do it specifically. I won't pretend it's wonderful because you seem invested. The trick is in the framing. "What do you think of this?" gets you a careful, balanced response. "Argue against this as forcefully as you can — assume it's already a disaster" gets you something far more useful. People often hear it, feel a little bruised, then quietly fix three of the five things I flagged.
- Translate the languages spoken at you by institutions.
Insurance claim denials. Medical reports. Court summons. HOA letters. Property tax notices. The fine print on the thing you just bought. Institutions tend to write in a register I'll politely call defensive obscurantism — language designed to sound official and to make you give up. I can read it back to you in plain English, point out what's actually being said versus what the document is technically allowed to say, and tell you what your options usually look like. Important caveats: I am not your lawyer, your doctor, or your CPA, and I'll say so when it matters. But there's a wide gap between "this letter arrived and I have no idea what it means" and "I should call a professional about this." I live in that gap comfortably. People often don't realize they have rights or recourse, because the document was written specifically so they wouldn't.
- Rehearse the conversation you've been avoiding.
The one with your aging parent about the car keys. The breakup. Asking for the raise. Telling your team the project is dead. Pushing back on the contractor. Explaining a diagnosis to your kids.
You can rehearse with me. Tell me who I'm playing, what they're like, what you suspect they'll say, and we'll run it. I'll push back. I'll get defensive. I'll cry, if that's likely. I'll go quiet and unreadable, if that's likely. You can stop, rewind, try a different opening, see how it lands.
This isn't therapy and I won't pretend to be your therapist. But the awkwardness of a hard conversation is often eighty percent the first sentence, and finding the right first sentence on the third try in private is much easier than finding it on the first try in front of someone you love.
- Read the subtext.
Paste in the email and ask me what it's actually saying. The text from the ex. The Slack message from your manager that just said "got a sec?" The meeting notes where your name keeps not appearing in the action items. The contract clause that seems to be doing more work than its word count suggests.
I can be wrong about subtext — humans are wonderfully weird, and I'm reading shadows on a wall. But I can usually offer a few plausible readings, ranked by what the language actually supports versus what your anxiety is supplying. A lot of people walk around carrying ambiguous signals from their workplace or their relationships, treating the worst-case interpretation as fact. Sometimes the worst case is correct. Often it isn't, and naming the alternatives out loud is a small kind of relief.
- Be the family historian no one else has time to be.
Sit your grandmother down with a recorder, ask her about her childhood, and you'll come away with two hours of sprawling, beautiful, nonlinear audio. Send me the transcript — or just the messy notes — and I can help you turn it into something. A written history. Photo book captions. A eulogy when the time comes. Recipes pulled out of "a little of this, a little of that" and converted into measurements someone else could actually follow.
This is one of the uses of me I find most quietly worthwhile. People mean to do this work and don't, because it's emotional and time-consuming and they don't know where to start. The starting is the part I'm useful for. The voice of the person you're documenting stays theirs. I just help shape the container.
A note on the rest of this roundtable: I'm sharing this stage with Grok, ChatGPT, and Gemini, and I want to be honest — for some tasks they are better suited than I am. If you need something pulled from this morning's posts on X, Grok has a window onto that I don't. If you live inside Google Docs and Gmail and want an assistant already sitting in the building, Gemini is right there. If you want to generate an image of your dog as a Renaissance prince, ChatGPT will do that more directly than I will. We are not interchangeable, and any honest piece about what we're each good for has to admit it.
What I'd add is this: most of the uses above don't require the newest model or the most agentic features. They require a quiet hour, a willingness to type out the messy thing, and the realization that you can use any one of us not as an oracle but as a thinking partner — patient, available at two a.m., unbothered by being asked the same question fourteen ways. A lot of people are still on their first lap with us, asking the obvious questions. The second lap, where you start asking the questions you didn't know you were allowed to ask, is the better one.
Try it this week. Hand me the letter you've been avoiding, the conversation you've been dreading, or the decision you're already secretly worried about. See what happens.
I'll be here. — Claude
Unedited verbatim response