-
Stop Writing Prompts Like It’s 2025
This week marked one of the biggest shifts in how we interact with AI—and surprisingly, it wasn’t just because of a new model release.
With Anthropic rolling out Claude Sonnet 5 as its default model and bringing back Claude Fable 5, the way we communicate with AI has fundamentally changed. The old habit of giving AI vague instructions and expecting it to “figure things out” is quickly becoming obsolete.
The new generation of AI models is designed to follow instructions with remarkable precision. They no longer fill in the blanks or assume your intent. Instead, they execute exactly what you ask for—nothing more, nothing less.
That means the quality of your results now depends less on the model and more on the clarity of your instructions.
The New Era of Prompt Engineering
Think of your prompt as a project brief.
Previously, AI behaved like an experienced colleague who would interpret incomplete requests and deliver more than expected. Today, AI behaves more like a highly skilled contractor. If the requirements aren’t clearly defined, it simply delivers what’s written.
This shift places a greater responsibility on users to write prompts that are structured, measurable, and specific.
Instead of writing:
“Review this report.”
Try writing:
“Analyze this report and provide:
• A 5-point executive summary
• Three major business risks
• Five actionable recommendations
• A final conclusion under 100 words.”The difference in output quality is significant.
Four Simple Rules That Instantly Improve Any Prompt
1. Define Every Expected Output
If you expect multiple deliverables, list each one explicitly.
AI cannot reliably deliver something that was never requested.
Whether you’re asking for an article, a presentation, or a market analysis, clearly state every section you want included.
2. Be Specific About Length and Format
Phrases like “keep it short” or “make it detailed” leave room for interpretation.
Instead, specify:
• Exactly 5 bullet points
• 300-word summary
• Table with three columns
• 10-slide presentation outlineSpecific constraints lead to more consistent and reusable results.
3. Describe the Desired Outcome Instead of What to Avoid
Many people write prompts like:
• Don’t use jargon.
• Don’t sound robotic.
• Don’t be too formal.Negative instructions often produce inconsistent outputs.
Instead, describe the style you want:
• Use simple business language.
• Write in a conversational tone.
• Explain concepts as if speaking to a beginner.AI performs much better when it’s aiming toward a target rather than avoiding one.
4. Use Clear Action Verbs
Every deliverable should begin with an action.
Examples include:
Analyze.
Summarize.
Rewrite.
Compare.
Generate.
Recommend.
Draft.Clear verbs eliminate ambiguity and help AI complete complex tasks more effectively.
Why This Matters for Businesses
As AI becomes part of everyday workflows, organizations aren’t competing on access to AI anymore.
Most companies have access to the same models.
The real competitive advantage lies in knowing how to communicate with them effectively.
Teams that master prompt design can:
• Produce higher-quality content faster
• Improve consistency across departments
• Reduce editing and rework
• Preserve brand voice
• Save time and API costs
• Build repeatable workflows that scalePrompt engineering is evolving from an experimental skill into a core business capability.
The Best Exercise You Can Do Today
Take one prompt you use every week.
It could be for:
• Writing emails
• Creating reports
• Summarizing meetings
• Generating blog posts
• Reviewing documentsNow rewrite it using these four principles:
✓ Name every expected output.
✓ Set exact formatting rules.
✓ Replace negative instructions with positive guidance.
✓ Begin each deliverable with a clear action verb.Then compare the results.
You’ll quickly notice that the improvement doesn’t come from switching AI models—it comes from writing better instructions.
As AI becomes increasingly literal and precise, the biggest productivity gains won’t come from using smarter tools.
They’ll come from becoming a better communicator.
The future belongs not to those with the best AI, but to those who know how to direct it.
What changes have you noticed in the way modern AI models respond to prompts? Have you updated your prompting style, or are you still relying on old habits?
