Elevating Your Coding Workflow with 'AI Context'
One of the most practical new AI prompting techniques comes from developer Maxi Contieri, who suggests treating pull request (PR) descriptions as training data for your AI and your team. The problem with many AI coding sessions is that the reasoning behind a solution is lost once the chat window closes. By adding an 'AI Context' block to your PRs, you create a permanent record of the 'why'.
The fix is simple. In every non-trivial PR description, include a dedicated section that details:
**AI Context:**
- **Tool & Model:** [e.g., Cursor with Claude Code]
- **Key Prompt:** [The exact prompt that unlocked the solution]
- **Failed Attempts:** [What the AI tried first that didn't work]
- **Manual Corrections:** [Any edits you made to the AI's final output]This practice turns every merged PR into a lesson that makes future AI sessions more effective. Your team, and any future AI agents, can learn from past successes and failures instead of starting from scratch every time.
Mastering Visuals: Prompts for Video and Photography
Advanced prompting is not limited to text and code. With the right instructions, you can direct AI models to create stunning and complex visual media. Here are two powerful examples for video animation and product photography.
Creating Video and Animation with Claude Design
You can use Claude's Opus 4.7 model to generate sprite-based animations and videos. The key is to provide a detailed prompt that outlines the style, content, and branding elements. Claude will typically ask clarifying questions about duration and aspect ratio before generating the video.
Here is a sample prompt to get started:
Create a sprite-based animation that introduces trivia about a [topic or brand]. Combine multiple animation styles, including engaging text animations, to enhance storytelling. Use colors, typography, and visual elements that align with the identity and feel of the [topic].Midjourney for High-End Product Photography
For static images, tools like Midjourney can produce photorealistic results that rival professional shoots. The secret lies in a highly descriptive prompt that specifies lighting, composition, mood, and even technical details like lens type and film look. This prompt for a skincare product is a masterclass in detail:
minimalist skincare product photography, editorial still life, an ambar spray bottle with no label, placed on a natural limestone stone pedestal, soft linen fabric backdrop in warm white and beige tones with natural folds, soft natural window light, side lighting, gentle organic shadows, airy and calm atmosphere, clean beauty aesthetic, modern organic style, australian skincare vibe, shot on 35mm film look, shallow depth of field, realistic photography, high-end beauty campaign, warm neutral color palette, ultra realistic, high detail, professional studio lighting --ar 2:3 --p 37bx8pbStrategic and Practical Prompting Frameworks
Beyond specific use cases, new frameworks for prompting are helping users get more reliable and structured results. These AI prompting techniques focus on how you format your request to the AI.
The Power of JSON Prompting
One trending technique is to write your prompts in code, specifically using JSON (JavaScript Object Notation). By structuring your request as a JSON object with clear keys and values, you can guide the AI to provide a response in the exact format you need. This is especially useful for tasks that require structured data output and can lead to drastically better results.
Defining Your Brand with a UVP Prompt
For business and marketing, a well-structured prompt can help you hone your messaging. This template asks the AI to act as a branding expert to help you clarify your Unique Value Proposition (UVP):
Prompt: Act as a positioning and branding expert. Here is my product or service: [describe clearly]. Here are my competitors: [list competitors]. Here is what my product currently claims to do: [current positioning]. I want you to craft a sharp and differentiated value proposition by identifying what makes this offer meaningfully different, why customers should care, and how it delivers value better or faster than alternatives. Validate the positioning logically and provide one primary value proposition and three secondary supporting statements.Building Custom Workflows and Dashboards
The most advanced use of prompting involves creating entire workflows and tools. One community member shared how they used the Claude platform to build a custom exercise tracking app, exporting the code to Bolt to create a functional web application. Another powerful workflow involves using Claude Cowork's Live Artifacts feature to build a daily command center. By prompting the AI to interview you about your work, KPIs, and connected apps, you can generate a live dashboard that pulls in data from Slack, email, and calendars, turning the chat interface into a personalized productivity hub.