Don't be a meat proxy: Why verbatim AI output adds no value

The meat proxy problem

Conclusion: Meat proxying—pasting AI output without reading or rephrasing—adds no value and burdens recipients.

The author observes that when a question is asked in Slack, a merge request, or a WhatsApp group, the reply often consists of a verbatim block of AI output prefixed with "Claude said:" or similar. This practice forces the recipient to read the AI’s raw response instead of receiving a concise, human‑crafted answer. The author notes that they have both done this and been on the receiving end, and that it provides no added value because the recipient could ask the AI directly and control the context.

Why verbatim AI output is burdensome

Conclusion: Verbatim AI output is verbose, jargon‑dense, and often contains plausible nonsense, forcing readers to spend extra effort to understand it.

The author describes AI output as extra effort to read, frequently verbose, containing plausible nonsense, and increasingly dense with jargon. As an example, they quote a sentence they received from Claude:

NATS control-plane events: stream leader election / R3 quorum re-form during pod churn.

The author states they had to look up almost every word to make sense of the sentence, illustrating the cognitive load placed on the reader.

How to add value when using AI

Conclusion: To add value, read the AI output, understand it, verify its correctness, and then write a response in your own words.

The author advises that prompting the AI is acceptable, but the output should not be forwarded unchanged. Instead, the user should read the AI’s response, comprehend it, validate its accuracy, and then produce a reply in their own language. This effort serves as a certificate that the prior steps were completed and provides genuine value to the conversation.

Code review as a meat proxy scenario

Conclusion: Using AI to generate code review feedback without personal examination turns the reviewer into a meat proxy and shifts responsibility to the AI.

In the context of code review, the author describes a pattern where a reviewer copies a ticket description into Claude Code, does not inspect the generated code, and then forwards any reviewer comments back into the AI for further iteration. The author argues that in this workflow the implementation work is performed by the reviewers using the AI, while the person forwarding the output acts merely as a meat proxy, contributing no original effort.

Community reactions: frustrations and suggestions

Conclusion: Commenters on Hacker News describe the practice as exhausting, suggest personal accountability, and propose ways to mitigate the issue.

Many commenters echoed the author’s frustration. @eddythompson80 wrote:

I deal with this all day long at work and it’s exhausting. People almost acting like no one has thought of it “I asked Claude what happened, and it spit out this 300 line response. Can you read it for me and see if it’s right?”

@mft_ noted a correlation between the behavior and perceived competence in their organization:

At a company I work for, I was recently discussing the need for an AI Code of Conduct plus a discussion - as this is exactly the sort of thing that started creeping in as soon as the corporate Claude account went live.

@sandeepkd observed irony in complaints about meat proxying:

From what I have seen personally over the last 8-12 months, I find this kind of complain ironical in most scenarios. Most of the people complaining this are comfortable with their own usage and distribution of LLM generated content and feel good about it, where as when they hear it from some one else it feels burdensome.

@gimili emphasized understanding the core idea comprehension:

For me what matters most is whether the person understood the core of the thing they are forwarding. Usually a wall of text is contraindicative of that. I want my team to be able to write the core idea on a whiteboard without any help before going ahead and sharing or implementing it.

@theletterf described a personal etiquette for using AI:

When someone asks me something and I use an LLM to research the answer (notice that I said 'research', not 'produce'), I frame the findings as something that I've investigated with, say, Claude. I also take the time to edit the response and read if myself at least twice, because I don't want a stochastic machine to take over my identity.

@felipeerias suggested a technical approach to reduce AI‑specific language:

One way to prevent obvious AI language from sneaking in text destined for other human beings is to ask the model to produce ASD-STE100 Simplified Technical English bullet points. This will result in a list of sentences that are clear and explanatory, easier to double-check, and convenient for the user to rewrite into a more readable format with a human voice.

@rockbruno highlighted the risk of treating AI as an independent authority:

The problem here is that the paradigm is AI-assisted development, but a LOT of people are treating it as "AI-independent" development instead, a.k.a "just send it to the agent and blindly trust whatever comes out of it", including deferring all responsibility / blame to the AI itself which is absolutely ridiculous.

These comments collectively confirm that meat proxying is perceived as a burden, that users should verify and rephrase AI output, and that clear communication of one’s own understanding is valued.