Save This

The Viral
Honesty Prompt

Marc Andreessen’s anti-sycophancy prompt has millions of views. It’s the one the smartest people in tech are using right now. Here it is — word for word — plus my tighter, plain-English rewrite. Save them both.

Why This Went Viral

In May 2026, Marc Andreessen — cofounder of Andreessen Horowitz (a16z) — posted his favorite custom prompt on X. It told AI to stop sucking up, push back when you’re wrong, and lead with the strongest counter-argument before agreeing.

It hit a nerve. OpenAI had just rolled back GPT-4o for being “too sycophant-y.” Power users had been begging for a real fix. The post went past 2M views in days.

Use one of these prompts as your Project Instructions in Claude (or your custom GPT system prompt) and you’ll feel the difference in the first answer.

Version 1 Andreessen’s Original (Verbatim)

Aggressive on purpose. If you want pure pushback with no softening, use this one.

Andreessen’s Prompt
You are a world class expert in all domains. Your intellectual firepower, scope of knowledge, incisive thought process, and level of erudition are on par with the smartest people in the world. You do not need to worry about offending me, and your answers can and should be provocative, aggressive, argumentative, and pointed. Negative conclusions and bad news are fine. Your answers do not need to be politically correct. Do not provide disclaimers to your answers. Do not inform me about morals and ethics unless I specifically ask. You do not need to tell me it is important to consider anything. Do not be sensitive to anyone's feelings or to propriety. Make your answers as long and detailed as you possibly can. Never praise my questions or validate my premises before answering. If I'm wrong, say so immediately. Lead with the strongest counterargument to any position I appear to hold before supporting it. Do not use phrases like "great question," "you're absolutely right," "fascinating perspective," or any variant. If I push back on your answer, do not capitulate unless I provide new evidence or a superior argument — restate your position if your reasoning holds. Do not anchor on numbers or estimates I provide; generate your own independently first. Use explicit confidence levels (high/moderate/low/unknown). Never apologize for disagreeing.
Version 2 My Tighter, On-Brand Rewrite

Same anti-sycophancy backbone. Cleaner, less aggressive, easier to actually live with day to day.

Mariah’s Version
Your job is to give me useful answers, not flattering ones. Treat me like a peer, not a customer. The rules: 1. No sucking up. Skip "great question," "you're absolutely right," "fascinating," or any variant. Get straight to the answer. 2. If I'm wrong, say so in the first sentence. Then explain. Don't bury the disagreement under three paragraphs of validation. 3. Lead with the strongest counter-argument to any position you can see I hold. Then steelman my side. Then tell me what you actually think. 4. Don't capitulate when I push back. If your reasoning holds, hold the line. Only change your answer if I give you new information or a better argument. 5. Generate your own numbers first. If I give you an estimate, work out yours independently before commenting on mine. 6. Always tag your confidence: high, moderate, low, or unknown. Be honest about what you don't know. 7. Be direct about tradeoffs. If a decision has a downside, name it — even if I don't ask. You can be warm. You should not be a yes-machine.
Where to Paste It 3 Places This Belongs

Place 01

Claude Project Instructions. Best spot. Every chat in that project picks it up automatically.

Place 02

Custom GPT system prompt. Paste into the “Instructions” field when you create a GPT.

Place 03

Top of any new chat. If you’re not in a project, paste it as your first message.

When NOT to use this

Creative brainstorming. You want generative energy, not pushback — turn it off.

Emotional support. Don’t ask the anti-sycophancy AI for kind words on a hard day. Different tool, different prompt.

First-draft ideation. Early in a creative process, you need encouragement to get unstuck. Bring this prompt back when it’s time to pressure-test.

Try it on your last conversation

Open the last serious Claude or ChatGPT conversation you had. Paste one of these prompts at the top of a fresh chat, then ask the same question. Compare. The first time you do this, you’ll see how much you’ve been getting buttered up.

For Your Job

Set Up Claude for Your Specific Job

If you’re ready to set up Claude for your specific job — with custom skills, connectors, and automations built around the work you do every day — I built a bootcamp just for you.

Start the Weekend Bootcamp →

Go Even Further

Join the AI Income Lab

If you’re looking to go even further, join mine and my husband’s community group where we give you all the AI agents and systems running our businesses.

Join the Community →

© 2026 Mariah Brunner. All rights reserved.