The Problem and the Fix
Without a skill
- You get long AI responses that are technically correct but useless for your specific situation.
- You can't tell whether a bad output is because of your prompt or because of AI's limits.
- You copy prompts from Reddit but they work for someone else's task, not yours.
With aidowith.me
- Apply 4 prompt elements - role, task, constraints, format - to any AI tool and get usable output faster.
- Build prompts on 15 real tasks from your own work, not generic examples.
- End with a personal prompt library that makes every future AI task faster.
Who Builds This With AI
Marketers
Content, campaigns, and briefs done in hours instead of days.
Sales & BizDev
Prep calls, draft outreach, research prospects in minutes.
Managers & Leads
Reports, presentations, and team comms handled faster.
How It Works
Audit your current prompts
You'll look at 3 prompts that gave you bad results and identify which of the 4 elements were missing. Most people are missing constraints and format. This step takes 10 minutes.
Rebuild with all 4 elements
You'll rewrite each prompt with role, task, constraints, and format - then compare outputs. The improvement is usually dramatic enough that you won't go back to bare prompts.
Build your prompt library
Each task adds a working prompt to your library. By step 15 you have 15 tested prompts and a fill-in-the-blank template for new tasks. The library lives in a doc you own and update.
Prompt AI Like Someone Who Knows What They're Doing
Practical Prompts at aidowith.me: 15 real tasks, 4-element framework, personal prompt library. 75 minutes.
Start This Skill →What You Walk Away With
Audit your current prompts
Rebuild with all 4 elements
Build your prompt library
End with a personal prompt library that makes every future AI task faster.
"I used to think I was bad at using AI. Turns out I just didn't know how to write a prompt. This route fixed that in one sitting."- Operations analyst, retail chain
Questions
Set the tone in the constraints section. Tell the model your audience, the register you want (casual, professional, direct), and what to avoid. The Practical Prompts route at aidowith.me shows you exactly how to do this on real work tasks so the output sounds like you, not like a press release. A tone line like 'write this as if explaining to a smart colleague - no jargon, no bullet overload' makes a significant difference in output quality.
Yes, arguably more. Newer models are more capable but also more literal - they do exactly what you say. Clear structure matters more, not less. The 4-element framework works on any model: GPT-4, Claude, Gemini, and whatever comes next. A well-structured prompt that worked on GPT-3.5 will work on any newer model, while a vague prompt still gets vague output regardless of how capable the model is.
Add a format constraint: 'Response: 3 bullet points, max 20 words each' or 'Answer in 2 sentences.' Put this at the end of your prompt. It's one of the first things covered in the Practical Prompts route at aidowith.me and it works immediately. Format constraints are also reusable - once you know the right format line for your output type, you add it to every prompt for that task and stop getting walls of text.