Prompting Craft
Stop getting generic AI drafts .
Three sentences of the actual situation (what has happened, what is at stake, why it isn’t routine) pull the model into your specific work. The single largest lever in prompt-writing, applied to your next non-trivial prompt.
No credit card required. Cancel anytime.
Why not buy a prompt-engineering course?
A search for "prompt engineering course" surfaces a hundred paid offerings. Coursera specializations, LinkedIn Learning paths, $499 LinkedIn "AI Prompt Engineer" bootcamps, $1,500 Udemy bundles, six-week corporate trainings. Many of them are genuinely useful. None of them will be a better use of two minutes than the single move this trail teaches.
The move is this: before you ask the AI for anything non-trivial, write three sentences of the actual situation above the ask. What has happened. What is at stake. Why this is not routine. Those three sentences carry more signal to the model than any list of adjectives, instructions, or persona descriptions.
Every prompting framework (CRISPE, RACE, the OpenAI "best practices," every paid course) eventually arrives at this point. We teach it first, in about 2 minutes, applied to a real prompt you were going to send anyway.
Questions
- Is my data private?
- Yes. Your prompts and AI responses are deletable, exportable, and only visible to you. We never expose individual prompts to admins, teammates, or Joring employees.
- How long should the backstory be?
- Three sentences is the right size. Long enough to carry the specific situation, short enough that you’ll actually write it. If you find yourself writing five or more, the prompt itself probably needs to be broken into parts.
- What if my situation is sensitive?
- Use the same level of detail you’d put in a Slack message to a trusted teammate. If a specific name, number, or fact would cross a line at your company, paraphrase it. The model needs the *shape* of the situation, not the legal disclosures.
- Does this work for code prompts too?
- Yes. The equivalent backstory for code is "what we’re building, what just stopped working, what we’ve already tried." Three sentences of project context turn a generic Stack Overflow answer into one written for your codebase.
- Does this work in ChatGPT, Claude, and Gemini?
- Yes, equally well. Adding situation context is the move that pays off in every modern LLM. They all reward specific context over generic instructions.
- What happens after the 14-day trial?
- Continue on Pro to keep real-time coaching, guided trails, and your exportable AI record. Or stay on Free with limited features. The trail progress you’ve already made doesn’t disappear.
Get drafts that could only have come from inside your work.
No credit card required. Cancel anytime.