HeyJaredAI Prompt Hackers
Join over 22,000 professionals, creators, and founders who want to use AI to think better, build faster, and stay in control. AI frameworks that sharpen your edge, not dull your thinking.
- Indexed issues, last 90 days
- 15
- Latest publication
- Sep 29, 2026
- Audience
- Checking…
- Earliest in this view
- Aug 27, 2026
Latest issues
Stop Losing Your Best AI Research in Old Chats (opens the original)
Read excerpt
Build a research notebook in ChatGPT that you can update and check as you go.You’ve read the report. You’ve asked ChatGPT to pull out the useful bits. The answer was good.Then another report arrives and changes the picture. The first answer is buried in an old chat. The second is in a new one. You now have to remember which finding came from which document and whether the two reports agree.Andrej Karpathy’s LLM Wiki proposal has a sensible answer: keep the original sources and
How to Stop AI Writing Sounding Generic (opens the original)
Read excerpt
7 constraints to add to your prompt files that eliminate robotic filler, corporate cliches and predictable generic frameworks.You wrote a good prompt. Role, audience, tone, word count, two examples. The output still came back sounding like a press release that had been through a compliance review.The reason is mechanical. A model fills every gap you leave with the most probable option, and probable means the phrasing everyone else’s output already contains. Your prompt described the shape of the
The prompt structure I stopped using after it made me lazier (opens the original)
Read excerpt
There’s a prompt structure almost everyone lands on eventually. Role, task, context, format, constraints. Feed it a topic, get back something that looks finished.It works. That’s the problem.After a year of using it for anything that mattered, the pattern showed up. Turning up to a client call with a document instead of a view. Somebody else’s reasoning, wearing your name on it.So it got demoted. Still fine for production work, banned for thinking work. What replaced it is a set of prompts that
Mort, Brian & Cooper (opens the original)
Read excerpt
<img alt="" class="sizing-normal" height="485" src="https://substackcdn.com/image/fetch/$s_!oFgZ!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fbec6bb35-b31b-421e-987c-b19903689498_2172x724.png"
Why your AI gives worse answers when the stakes are higher (and what to do about it) (opens the original)
Read excerpt
Ask an AI to name a decent restaurant in York and you’ll get a useful answer. Ask it whether you should drop the client who pays 40% of your income, and you’ll get something that reads just as confidently and is worth a great deal less.This isn’t a feeling. Researchers at Stanford ran 11 models across thousands of advice scenarios in March and found the models endorsed the user’s position 49% more often than a human adviser would. Where the user’s described behaviour was actually harmful, the mo
Publishing over time
Last 90 days. Choose a month to open its work.
Recurring subjects
Named in the text we hold. One piece can cover several.
Audience
No verified audience measurement yet.
About this data
Counts cover the work we have indexed. Tone needs enough text and a confident classification. Excerpts and episode notes are not full articles or transcripts.
Identity or attribution wrong? Suggest a correction.