My experiments with AI

🌟 Welcome to Intuit Path – my weekly blog where I explore the evolving world of artificial intelligence through personal insights, hands-on experiments, and practical reflections from my daily life as a developer.

🧠 Idea or Spark

While commuting to work, I thought: What if I record everything I’m learning about AI each day? Years from now, I’ll be able to look back and see how much the field—and my own thinking—has evolved.

🗣️ Context or Backstory

This idea came to me as I was driving and listening to the Hard Fork podcast episode featuring Demis Hassabis, CEO of Google DeepMind. It was a discussion on the future of AI and recent announcements at Google I/O. I got so inspired that I started talking to ChatGPT-4.0 in the car, hoping it would log and format everything. But when I got to work, the conversation history was gone—and with it, the format I wanted for my blog post. That frustration ended up reinforcing the need to start this very journal.

I’ve been using AI nearly every day—both personally and professionally—and often switch between tools depending on what I need. Today, I tried using both ChatGPT and Grok to kickstart a journaling habit. Each had its own strengths and weaknesses.

ChatGPT helped organize my ideas but lost my chat history. Grok retained memory but repeated context excessively, making it hard to build on ideas naturally. I realized I need an AI journaling tool that supports voice input and saves formatted entries reliably.

🤖 AI Tool Comparison

ChatGPT 4.0 – The versatile powerhouse

• Planning & Organization: Excellent for complex projects like travel planning, coding curricula, piano lesson customization, and garden design planning.
  • Since it has real time web access, I am able to provide it links and ask it to give me a summary or comparison of options. This helps Since it has real-time web access, I can provide links and ask for summaries or comparisons of options. This helps me digest data much quicker and make faster, data-driven decisions
  • For example, when I was researching accommodation options for a family get together, I fed it my requirement and short listed property links and ChatGPT generated a perfect comparison table that streamlined my decision-making.
• Visualizations: Game-changer for my garden project – created realistic overlays on my original photos showing proposed plantings and design changes. These AI-generated previews made decision-making so much easier and more confident
• Journaling: Voice mode enables natural, conversational entries with excellent formatting and organization. Access to external tools gives it a major edge over competitors
  • Bottom line: My top choice for journaling and complex planning tasks

Grok 3 – The conversationalist with limitations

• Strengths: Engaging dialogue style with excellent context retention across long conversations
  • Charlotte’s Punta Cana memoir was a successful journaling experiment that helped capture her experiences with vivid details and personal reflections. Grok 3 assisted with the creative writing, preserving her authentic 9-year-old voice. (Read it here).
• Major weakness: Tends to repeat entire contexts in longer sessions, making it hard to build progressively on ideas
  • To overcome this, I copy pasted some of the content I had added here into ChatGPT to get the full document formatted. I also used Google docs to keep my latest version which ChatGPT could access to get the whole memoir instead of pulling it from memory
• Critical limitation: No real-time web access – when I asked for a Hard Fork podcast summary, it pulled outdated content from February instead of the current episode I’d just heard
• Bottom line: Great for sustained conversation, but limited by information gaps and repetitive output

Key Takeaway: Each tool has its sweet spot – ChatGPT for structured work and journaling, Grok for conversational exploration (when context repetition isn’t an issue). The trick is matching the right AI to your specific task.

🔗 AI News that caught my eye

• Anthropic Claude 4 Announcement: Anthropic unveiled Claude 4 with a significant pricing jump, reflecting its ambition in the AGI race. Read the announcement
  • Opus 4 at $15/$75 per million tokens (input/output) and Sonnet 4 at $3/$15.
• Hard Fork Podcast – Google DeepMind CEO Demis Hassabis on Living in an AI Future: A thoughtful conversation on responsible AI and Google’s vision for Gemini. Listen here
  • Artificial General Intelligence (AGI)—highlighted in the Hard Fork podcast—refers to an AI system with human-like cognitive abilities. It raises the question: how soon might AGI become part of everyday life, and what will that change look like for us?
  • I wanted to checkout Gemini immediately but look at their pricing model for 2025
    • Google AI Pro – $20/month: Access to Gemini 2.5, Deep Research, and Veo 2 video generation.
    • Google AI Ultra – $125/month: Includes highest feature limits, exclusive access to Gemini 2.5 Pro Deep Think (most advanced reasoning model, coming soon), and Veo 3 video generation.
• AWS Transform General Availability Launch: A proud moment—our team’s product, AWS Transform, is now generally available! Reflects months of work on scalable, AI-powered transformation tools. See the announcement
  • AWS Transform for Mainframe: Another post that highlights our work in bringing modern AI-powered transformations to legacy systems. I was intricately involved in this initiative and it’s fulfilling to see its impact unfold. View post
  • AWS Transform Web App Video by Matt Garman: Every time I see this video it fills me with pride—I developed the core part of the JobsUI page shown. It’s incredible to see our team’s work celebrated so publicly. Watch video
• Amazon Q updates
  • Amazon Q GitHub Launch: Exciting to see Amazon Q available for GitHub. As a daily user, I can vouch for how much it streamlines development and empowers productivity. View post
  • Amazon Q used to summarize PR’s – This is another capability I’ve been able to leverage through Amazon Q for a while now that is available for GitHub. It demonstrates an emerging technical direction in model usage. View post
  • Amazon Q CLI with MCP: The release of the Q CLI with Model Context Protocol support for external users is a huge step forward. I’ve been using it every day to accelerate development, generate documentation, and boost code quality. View post
• CodeCatalyst Idea2PR Highlight: Seeing our earlier work on Idea2PR recognized in the AI space made me proud. It reminded me of the early steps we took in integrating DevOps and AI thinking. View post

🌱 Future Ideas to Explore

• Teach Philip and Charlotte to journal with AI this summer.
  • It would be great to create more lasting family memories with tech-powered storytelling.
• Summer Planning – I want to use AI to create a full summer program for the kids, including an engaging coding curriculum tailored to their interests.
• New AI Tools I Want to Try: GitHub Copilot, Claude Code, OpenAI Codex, Google Jules, Cline, Sourcegraph Amp, Cognition Labs Devin – all promising in the coding assistant and AI development arena.

📘 Glossary / Fun Facts

• AGI – Human-level AI with general reasoning. Also known as human-level AI, it refers to hypothetical AI that possesses the ability to understand, learn, and apply intelligence across a wide range of tasks at a human level.
• MCP – Protocol for context handoff between AI models.
• Agentic AI – Autonomous AI with planning and decision-making skills.
• Voice-first input – Ideal for journaling on the go.