Comments on: My Company Put Me on an Extreme AI Token Diet for 2 Months. The Result Was Design Rot. https://www.designwhine.com/my-company-put-me-on-an-extreme-ai-token-diet/ Tech Through A Design Lens Tue, 23 Jun 2026 08:11:57 +0000 hourly 1 https://wordpress.org/?v=7.1.4 By: Justin C. https://www.designwhine.com/my-company-put-me-on-an-extreme-ai-token-diet/#comment-236 Tue, 23 Jun 2026 08:11:57 +0000 https://www.designwhine.com/?p=15941#comment-236 In reply to Oliver.

As a designer I sure miss the “manual” days

]]> By: Mathieu Côté https://www.designwhine.com/my-company-put-me-on-an-extreme-ai-token-diet/#comment-235 Mon, 22 Jun 2026 15:16:42 +0000 https://www.designwhine.com/?p=15941#comment-235 On: "To conserve tokens, teams often stretch single, massive chat threads across days or feed models incomplete context snippets."

I use multiple tools, all the research goes in ChatGPT, the synthesis goes to Magicpatterns and overall just being careful about using longer prompts, less back-and-forth, and trying not to waste tokens on stuff that probably didn’t need AI in the first place…

]]> By: Katherine Ashwell https://www.designwhine.com/my-company-put-me-on-an-extreme-ai-token-diet/#comment-234 Mon, 22 Jun 2026 15:12:19 +0000 https://www.designwhine.com/?p=15941#comment-234 On: "When you place a meter on that experimentation, you force the designer straight into safe, predictive convergence. It is death by a thousand efficiency metrics."

Honestly, I can relate to it. And this is not because the tools are bad (well in most cases), but because once you start watching every output like it costs you something, you stop experimenting and start settling.

]]> By: Oliver https://www.designwhine.com/my-company-put-me-on-an-extreme-ai-token-diet/#comment-233 Mon, 22 Jun 2026 15:08:09 +0000 https://www.designwhine.com/?p=15941#comment-233 On: "When you place a meter on that experimentation, you force the designer straight into safe, predictive convergence. It is death by a thousand efficiency metrics."

Honestly yeah, the ‘save tokens’ advice often turns into weird workflow hacks that just slow everyone down. If the model needs all this micromanagement, maybe we’re not ready for the expensive AI yet.

]]>