NinjaAI.com
Excerpts from "Combining AI Tools for Better Results : Social Media Examiner" by Michael Stelzner (July 22, 2025)
This briefing document summarizes key strategies and insights from Michael Stelzner's "Combining AI Tools for Better Results" article, co-created with Grace Leung, focusing on optimizing AI usage for marketing. The central theme revolves around a strategic, multi-model approach to AI, moving beyond reliance on a single tool like ChatGPT to leverage the distinct strengths of various platforms for enhanced performance and capabilities.
1. The Imperative of a Multi-Tool AI Approach
Key Idea: Limiting oneself to a single AI tool, even a popular one like ChatGPT, means "missing significant opportunities for enhanced performance and capabilities." Different AI models are trained differently, leading to fundamentally distinct capabilities and strengths.
- ChatGPT: Primarily a text-based model.
- Gemini: Trained multimodally from inception, capable of handling "text, images, audio, and video inputs simultaneously." Excels at data-heavy analysis and deep research.
- Claude: Focused on "linguistic sophistication and human-centered ethical considerations." Strong for refined analysis, strategic interpretation, and storytelling.
- Perplexity: Excels at rapid, real-time internet research with cited sources.
- NotebookLM: Unique for internal research, focusing exclusively on user-provided sources, minimizing hallucination, and maximizing relevance.
Actionable Insight: Marketers need "hands-on experience to determine which tools work best for your recurring tasks" rather than relying solely on general recommendations.
2. Establishing a Strategic Framework for Multi-Tool AI Workflows
Key Idea: A structured approach is essential for testing and implementing multiple AI platforms effectively.
a. AI Tool Testing Framework:
- Identify Recurring Tasks: Focus on "your most frequent, recurring tasks" (e.g., strategy development, content creation, data analysis, research).
- Standardized Prompts: Design a "standardized prompt that you can test across different models simultaneously" for objective comparison.
- Evaluation Factors:Speed: How quickly each tool responds.
- Accuracy: Checking for "bias or inaccurate information."
- Tone and Presentation Style: Significant variations exist.
b. Testing for Bias and Accuracy:
- Factual Questions: Ask questions about topics you know well to "spot potential biases or knowledge gaps."
- Brand/Industry Overviews: Request a comprehensive overview of your brand or industry to see "which model provides the most accurate and complete information."
- Chain-of-Thought Analysis: Examine the model's "reasoning process" to understand how it breaks down complex tasks.
c. Mapping Workflows:
- Strategic Leverage: After testing, "map each part of your systematic workflow to the tool that strategically leverages these capabilities."
- Example Workflow: "Perplexity for initial research, Gemini for data-heavy analysis, and Claude for refined strategic thinking and presentation."
- Goal: Develop a clear understanding of "which tool serves each part of your process most effectively and then build repeatable workflows around these insights."