The AI Arsenal
A Practical Guide to Claude, ChatGPT, Gemini, and Beyond
Now that we've built your prompt library, it's time to understand the AIs that will bring those prompts to life. If you're still using just one AI for everything, you're leaving capability on the table. Each AI has its strengths, and knowing when to utilize which AI can significantly enhance your results. I don't pretend to be an expert on this; you should do your research. However, I am sharing my personal perspective on these AIs here.
This isn't about finding the "best" AI, it's about building an arsenal where each AI serves its purpose. Think of it like having different specialists on your team: you wouldn't ask your accountant to design your logo, and you shouldn't ask every AI to do the same tasks.
Here is my take...
The Big Three: General Purpose AI
Claude: The Thoughtful Writer
Claude excels at crafting nuanced, well-written content that demands depth and consideration.
What Claude does best:
Long-form writing with sophisticated structure
Complex analysis that requires multiple perspectives
Detailed explanations with proper context
Content that needs a more human, conversational tone
Tasks requiring careful reasoning and ethical consideration
Writing style: Comprehensive, thoughtful, and detailed. Claude takes time to explore ideas fully and provides context for recommendations.
Best for: Blog posts, detailed reports, complex problem-solving, creative writing, and content that needs to sound natural and engaging.
Current models: Claude 4 Sonnet (Thinking), Claude 4 Opus for the most complex tasks.
ChatGPT: The Efficient Executor
ChatGPT is direct, concise, and gets to the point quickly.
What ChatGPT does best:
Quick, actionable responses
Straightforward task execution
Rapid brainstorming and ideation
Simple coding tasks and debugging
Tasks where brevity and speed matter
Writing style: Direct, concise, and practical. Gets straight to the answer without excessive elaboration.
Best for: Quick answers, rapid prototyping, simple automations, when you need results fast.
Current models: GPT-4o, GPT-4o mini for simpler tasks.
Gemini: The Detailed Processor
Gemini provides extensive detail and understandably excels at integration with Google services.
What Gemini does best:
Thorough analysis with detailed reasoning
Integration with Google Workspace (Docs, Sheets, Gmail)
Complex research tasks requiring deep thinking (Pro)
Processing large amounts of information
Tasks requiring step-by-step reasoning
Writing style: Comprehensive and methodical, often longer than necessary but shows its work.
Key versions:
Gemini Flash: Faster, good for everyday tasks
Gemini Pro: Better reasoning, excellent for coding and complex analysis
Best suited for: Research, Google Workspace integration, and tasks that require detailed explanations of reasoning.
Memory Across Chats
I am particularly interested in whether models can retain information across multiple chats, so here are the results:
One of the most practical differences between AIs is how they handle memory. Claude has historically been session-based, meaning once a chat ended, context was lost. Newer versions allow limited memory depending on the plan, but its long-term recall still feels less reliable compared to others. ChatGPT, by contrast, has the most developed memory system today. When enabled, it can remember details across conversations—like your project goals or past tasks—making it particularly strong for ongoing workflows where continuity matters. Gemini takes a different route: instead of emphasizing long-term memory inside the chat, it leans on its integration with Google Workspace and Google One. This gives it easy access to your docs, calendar, and drive, with clear user controls over what’s shared. In short, Claude is best when you want clean slates, ChatGPT excels at remembering evolving tasks, and Gemini fits if your work already lives in Google’s ecosystem.
*Update: Starting Oct 23, 2026, memory is rolling out to all Claude paid users. What makes it even more unique is the feature of allowing user to bring in saved memory data from another account or previous session, even an exported memory from ChatGPT can be uploaded to Claude.
Specialized: Coding
Windsurf: The Collaborative Coding Partner
Windsurf stands out as an AI-native IDE that transforms how you approach software development. (Disclaimer: This is what I used the most and am most familiar with.)
What makes Windsurf special:
Context-aware coding across entire projects (and I find it performs well across multiple projects too)
Collaborative editing that feels natural—the planning mode is nice
Multiple model access in one interface
Understanding of project structure and dependencies
Excellent at refactoring and code improvement
Note: Toggling between the Windsurf cascade and the command line can sometimes be challenging. I tend to stay within the Windsurf cascade.
Personal experience: I've found Windsurf particularly strong at understanding project context and making suggestions that actually fit your existing codebase architecture. It offers several models:
Claude 4 Sonnet (Thinking): Daily driver for most coding tasks
Claude 4 Opus: For complex project-wide planning (use sparingly - 5x more expensive)
Gemini 2.5 Pro: Great for documentation, cheaper alternative
OpenAI o3: Best for planning and project analysis, not day-to-day coding
Claude Code: Command Line Powerhouse
Claude Code brings AI assistance directly to your terminal. Best for: Developers who live in the terminal and want AI assistance without leaving their command-line workflow.
Terminal-native workflow integration
Command-line task automation
Git integration and code review
Direct file manipulation and project management
Maintaining context across development sessions
GitHub Copilot: The Code Completion Specialist
GitHub Copilot is the main AI coding assistant integrated with both Visual Studio and VS Code. Best for: Code completion, generating common patterns, working with familiar frameworks.
Real-time code suggestions and completions
Pattern recognition from vast code repositories
Boilerplate generation
Language-specific best practices
Specialized: Image Generation
In my line of work, I work heavily with text and code. I don't have personal experience with image generation AIs. The information below is generated by Claude and verified by ChatGPT and Gemini.
Precise and Prompt-Responsive
Strengths:
Excellent at following detailed text prompts
Good at incorporating text into images
Reliable and consistent results
Well-integrated with ChatGPT
Artistic and Stylistic
Strengths:
Superior artistic quality and aesthetics
Great for creative and conceptual work
Strong community and style references
Excellent for marketing and brand imagery
Commercial-Ready
Strengths:
Commercially safe (trained on licensed content)
Great integration with Adobe Creative Suite
Good for professional and business use
Reliable for consistent brand work
Specialized: Audio Generation
Shoutout to ElevenLabs for their effort in paying voice actors/actresses when generating voice with cloned voices, a responsible approach to AI audio generation.
The Multi AI Strategy
Here's how I actually use these AI together. For each activity, AI doesn't remove the human activity, such as ideation, editing, and decision-making. This list only explains the role and usage of AI.
For blog writing (other than the human input and editing):
Brainstorm with ChatGPT and Claude - Get quick ideas and outlines based on my initial ideas
Develop with Claude - Write the detailed, nuanced content that closely align with my writing style
Fact-check with ChatGPT and/or Gemini - Verify information and add research depth
Cross-check and request feedback on everything - Use each AI to review the others' work, i find this the most beneficial step, all of them bring in different perspectives
For coding projects:
Plan in Windsurf with Claude 4 Sonnet, use Claude 4 Opus sparingly, Claude 4 Sonnet is solid and serves the purpose well. Claude 4 Opus is an extremely capable model considered by most to be the best currently available for coding. Use sparingly for highly complex and "project wide" planning tasks. Extremely expensive (5x Claude 4 Sonnet)
Daily development with Claude 4 Sonnet - Most implementation work. I would recommend this to be everyone's go-to model except for simple tasks. Overkill for mundane tasks.
Critical Success Factors: the 80/20 Rule
The first thing I do when I can is turn off auto-execution.
This is crucial: Read everything and make human decisions. AIs are collaborators, not replacements for your judgment. I catch errors, provide suggestions for improvement, guide “my AI” where to look for solutions, correct errors, and frequently combine ideas from multiple AIs.
The first 80% of most tasks happen incredibly fast with AI. The last 20% (to be honest, maybe only the last 5%), especially in coding, takes the longest. Plan for this. The AI gets you to "good" quickly, but "great" still requires human refinement. Back up your work so that you can ROLL BACK.
Fun fact: In my experience with Windsurf and other coding IDEs, you'll get a working solution in minutes, but perfecting it for production is where you'll spend most of your time. Once you reach 80%, my experience is to make micro changes and be very specific. After working with Windsurf for a while, I have gained experience in identifying errors and specifying modifications, as well as verifying and re-verifying progress.
One of our developers jokingly said, "At launch, Sonnet 3.7 was known as the 'I fixed the typo and also rewrote your entire app in Rust you're welcome' model.” And it certainly feels very true.
On top of that, if you are utilizing Cloudflare Workers (headless), it can totally ignore your original local UI and build an entirely different look and feel interface just because the original was "too large". You will need to spend endless time convincing your AI and tweaking it.
Model Selection Matters
Each platform offers tiers that balance performance with price. Think of it like this:
Claude – Use Opus sparingly for complex reasoning; stick with Sonnet as your reliable daily driver.
ChatGPT – The Plus plan unlocks GPT-4o, which is generally enough for most ops workflows.
Gemini – The most cost-effective if you’re already in the Google ecosystem, with the added benefit of workspace integration.
The key is matching the model to the task: don’t overspend on power when speed and affordability will do. More powerful doesn't always mean better for your specific task.
AI API Pricing (as of Sep 10, 2025)
For more information about different AI products, view
[ 📥 AI Products Quick Reference ]
Building Your AI Workflow
Sometimes you don't have a choice of which AI to use, as your company might decide that for you. When you can, start with one primary AI that matches your most common use case. Then gradually add others:
Identify your primary use case (writing, coding, analysis)
Choose your daily driver based on your needs and working style
Add complementary AIs for specific tasks
Develop cross-checking habits - use different AIs to verify important work
Track what works - note which AIs combinations give you the best results
Moving Beyond Level 1
Most people use AI for simple content creation—the "write me a blog post" level. The real power comes from combining multiple AIs for different parts of complex workflows. Using AI for analysis and strategy, not just for execution. Build a system where AI works together. The goal isn't to replace your thinking—it's to amplify it. Each AI option in your arsenal should make you more capable, not more dependent.
Getting Started
Audit your current usage - What are you using AI for right now?
Identify gaps - Where could different AIs serve you better?
Experiment systematically - Try the same task with different AIs
Build gradual habits - Add one new AI at a time to your workflow
Share learnings - The AI landscape changes fast; community knowledge helps everyone
Remember: The best AI is the one that fits your workflow and thinking style. Start with what works, then expand strategically. Your prompt library from our previous post will work across all these AIs—now you just need to know which one to use, and when.
Build your arsenal thoughtfully, and you'll be ready for whatever comes next.