loudog.uno

AI discourse is exhausting. Half of Twitter thinks AI will solve climate change and cure cancer. The other half thinks it's a grift that will steal everyone's job and ruin art forever.

Meanwhile, I'm over here using AI to... rewrite awkward emails and summarize meeting notes.

The reality of AI in 2025 is way less dramatic than the discourse suggests.

1. Deep Research (Perplexity/ChatGPT)

This is the killer app right now. I can throw a complex research question at Perplexity or ChatGPT's Deep Research mode, and 10 minutes later get back a comprehensive report with sources.

Real use case: Planning a trip to Japan. Instead of spending hours googling, I asked:

"Plan a 10-day itinerary for Tokyo and Kyoto in April, optimized for food and architecture. Budget: $150/day. Prefer small local spots over tourist traps."

Got back a detailed day-by-day plan with neighborhood recommendations, specific restaurants, transportation tips, and budget breakdown. Would've taken me 5-6 hours. Took AI 10 minutes.

2. Email Alchemy

I write emails like I talk: stream of consciousness, overly casual, way too many em dashes.

AI is my email translator.

Brain dump → Feed to ChatGPT with "Make this professional but not stiff" → Get back polished email.

I save 10-15 minutes per professional email not agonizing over tone.

3. Code Completion (GitHub Copilot / Cursor)

It's not writing full programs for me. But it IS:

Filling in boilerplate: I type a comment, it generates the implementation
Explaining unfamiliar code: Highlight gnarly SQL query, ask "what does this do?"
Converting between languages: I write Python, get back TypeScript
Writing tests: I write the function, Copilot writes the test cases

Important: You need to understand code well enough to review what AI generates. It's a tool, not a replacement for knowledge.

4. Meeting Notes (Otter.ai / Fireflies)

AI transcription services solve the "take notes OR pay attention" problem.

Otter.ai joins my Zoom calls, transcribes everything, identifies speakers, and generates a summary.

After the meeting, I get full transcript, key points, action items, and searchable archive.

5. Image Generation (Midjourney / DALL-E)

I can't draw. But I can describe what I want, and AI can generate it.

Use cases:

Blog header images
Presentation visuals
Concept mockups for design ideas
Placeholder art for projects

6. Learning New Topics (ChatGPT as Tutor)

ChatGPT is an incredibly patient teacher.

You can have a full Socratic dialogue. It adapts to your understanding level. It doesn't get frustrated when you ask the same question three different ways.

This is how I learn now. Way more effective than random blog posts.

7. Idea Synthesis (Combining Notes)

I keep a personal wiki in Obsidian. Hundreds of notes on different topics.

I can dump 5-10 notes into ChatGPT and ask: "Synthesize these notes into a unified framework."

It identifies common themes, spots contradictions, and creates a coherent synthesis.

8. Translation for Real-World Use

AI chat models are significantly better than traditional translation for nuanced translation.

I can ask: "Translate this email to Spanish, formal business tone, Mexico context not Spain"

It handles context and tone in ways Google Translate doesn't.

9. Brainstorming Partner

I use AI for rubber-duck debugging my ideas.

Writing an article but stuck on the opening? Dump thoughts into ChatGPT and ask: "What's the interesting angle here?"

It bounces ideas back. Suggests framings I hadn't considered. Points out logical gaps.

10. Automating Boring Text Work

The least sexy but most time-saving use: text manipulation.

Examples:

"Extract all email addresses from this document"
"Convert this table to CSV format"
"Remove all URLs and formatting from this text"
"Create a summary bullet list from these meeting minutes"

I save probably 2-3 hours a week on this kind of stuff.

The Meta-Lesson

Notice what's NOT on this list: "Write my novel" or "Do my job for me."

AI is excellent at:

Accelerating tedious work
Providing first drafts
Explaining complex topics
Synthesizing information
Generating ideas to react to

AI is bad at:

Original creative vision
Understanding nuanced context
Making judgment calls
Anything requiring taste

The AI Stack I Actually Use

ChatGPT Plus ($20/month): Daily driver for text work
Perplexity Pro ($20/month): Research and fact-checking
Cursor ($20/month): Code editor with AI
Otter.ai ($17/month): Meeting transcription
Midjourney ($30/month): Image generation

Total: ~$110/month
Saves me: Easily 10-15 hours/week
ROI: Absolutely worth it

The hype is annoying. The doomerism is exhausting. But the tools? The tools are genuinely useful.

Start small. Pick one thing from this list and try it for a week. I bet you'll find it hard to go back.