Key Concepts
- Gemini 2.5 Pro Experimental: Google's latest, free AI model with strong reasoning and coding capabilities.
- DeepSeek V3.1: A free AI model known for content creation and humanization.
- AI Agents: Automated systems built using AI models to perform specific tasks.
- NA10: A platform for building AI agents, free for local or self-hosting.
- Open Router: A platform providing free API keys for various AI models, including Gemini 2.5 Pro Experimental and DeepSeek.
- Root Code/Client (Visual Studio Code): Code editors used for building and testing code generated by AI models.
- Make.com: A platform for creating automated workflows and integrations with AI models.
- API (Application Programming Interface): A set of rules and specifications that software programs can follow to communicate with each other.
- MCP (Multi-Control Point) Agents: Advanced AI agents with multiple control points for complex tasks.
Gemini 2.5 Pro Experimental Overview
- Announcement: Released on March 25th as Google's most intelligent model.
- Access: Available for free at a studio.google.com. Select "Gemini 2.5 Pro experimental" in the top right.
- Capabilities: Excels in reasoning and coding.
- Benchmarks:
- Leads on common benchmarks, including LM Arena.
- Outperforms previous Gemini models and competitors like GPT-4, Claude 3, and Grok.
- Dominates in humanities, GPQA diamond, mathematics, science, reasoning, and knowledge.
- Coding Example: Used to create a pixelated dinosaur endless runner game with one prompt.
- Usage: Can be used directly in the Gemini chat (advanced mode).
- Speed: Responds quickly.
- Reaction: Sundar Pichai (Google CEO) highlighted the dinosaur game.
Coding with Gemini 2.5 Pro Experimental
- Process:
- Come up with an idea.
- Give Gemini the code.
- Follow the instructions to run the code.
- Example: Creating a Pixel Dino Run game.
- Power: Enables rapid idea-to-execution.
Testing Gemini 2.5 Pro Experimental
- SEO Agency Landing Page: Basic prompt tested.
- Video Transcript to Blog: Used to create a 1,500-word article with headlines and links.
- SEO Cost Calculator:
- Prompted to create a sleek, modern design in HTML.
- Output tested on liveweave.com.
- Compared to Claude's output.
- Claude's version was preferred for its more complex calculator and UI.
Free API Access
- Open Router: Provides free API keys for Gemini Pro Experimental 2.5.
- Client/Root Code:
- Visual Studio Code extensions for coding.
- Open Router API key can be used in Root Code.
- Google's API key doesn't have 2.5 inside.
Building AI Agents with Gemini 2.5 Pro Experimental
- NA10:
- Free platform for building AI agents (if self-hosted).
- Steps:
- Sign into NA10.
- Add a new "chat trigger" step.
- Add an AI agent.
- Select "Google Gemini chat model" and "2.5 Pro Experimental."
- Get a free API key from a studio.google.com and plug it into the credentials section.
- Test the connection.
- Ask a question.
- Memory: Can add memory to remember previous interactions.
- Tools: Can connect to various apps (e.g., Gmail, Discord, GitHub).
- Gmail Automation:
- AI can figure out the "to," "subject," and "message" fields.
- Example: Sending an invitation email.
- WordPress Integration:
- Can create blog posts automatically.
- Example: Creating an SEO-optimized article.
- Prompt: "You're a helpful SEO assistant writing blog content based on my needs."
- Can specify HTML format.
- MCP Agents:
- Advanced agents with multiple control points.
- Resources available in the AI Profit Boardroom.
- Importing AI Agents:
- Easy to import pre-built agents into NA10.
Gemini 2.5 Pro Experimental vs. DeepSeek V3.1
- Coding a Game:
- Gemini 2.5 Pro was faster and produced a more addictive game.
- DeepSeek's game had design issues and broke easily.
- Content Creation:
- DeepSeek V3.1 produced more humanized content.
- Gemini 2.5 Pro's titles were average.
- Reasoning Challenge:
- Gemini 2.5 Pro recognized the impracticality of picking apples in winter.
- DeepSeek V3.1 focused on crossing the frozen river without questioning the premise.
- API Testing:
- DeepSeek's API key worked in Client, while Gemini 2.5 Pro's did not.
- Gemini 2.5 Pro worked inside Root Code.
- Gemini 2.5 Pro's API request got exhausted.
- Open Source:
- DeepSeek models are open source and can be hosted locally.
Building with DeepSeek V3.1
- Access: Available for free at deepseeek.com.
- Game Development:
- Used to create a captivating endless runner game.
- Code pasted into editor.p5js.org.
- Canvas Generator:
- huggingface.co/spaces/akhaaliq/nyhat
- Used to create a to-do list and an SEO cost calculator.
- NA10 Integration:
- AI agent built in 36 seconds.
- Email agent built in 47 seconds.
- Visual Studio Code (Root Code):
- Used to automate website creation.
- Free API key from Open Router.
- Make.com:
- Used to build workflows and automations.
- Can connect to Google Sheets and WordPress.
AI Profit Boardroom
- Content: Prompts, tips, workflows, AI agents, templates.
- Focus: Making more money and saving time with AI.
- Community: 543 members for questions and help.
- Weekly Q&A: Live calls for asking questions.
- Classroom: Crash course on AI systems, SAPs, workflows.
- Pricing: Prices are increasing at the end of the month.
Additional Offers
- Free One-to-One SEO Strategy Session: Learn how to take websites from zero to 145,000 visits a month.
- Free Link Building Acceleration Session: Get a custom-tailored link-building plan.
Conclusion
The video provides a comprehensive overview of the new Google Gemini 2.5 Pro Experimental and DeepSeek V3.1 AI models, demonstrating their capabilities and providing practical guidance on how to use them for various tasks such as coding, content creation, and automation. It highlights the strengths and weaknesses of each model, offering valuable insights for users looking to leverage AI in their workflows. The video also promotes the AI Profit Boardroom as a resource for accessing advanced AI tools, templates, and community support.
AI summaries can miss context or contain errors. Check important details against the original video.





