POWER-USER PLAYBOOK
The AI power-user stack (top tools, by use case) ⚡
Master this high-ROI AI workflow: This is my AI power user stack: the exact tools I open by use case, and what each one is actually for.
Master this high-ROI AI workflow: This is my AI power user stack: the exact tools I open by use case, and what each one is actually for.
The real unlock of AI agents is not full autonomy, but human-in-the-loop systems design. Structure your agent inputs with explicit constraints, negative prompts, and automated test checkpoints. When building tools, keep token consumption lean by caching system prompts and isolating tasks into specialized sub-agents.
The quick reference
| What I'm trying to do | The tool I open |
|---|---|
| Deep thinking + strategy | Claude Opus 4.8 |
| Design | Claude Design |
| Image generation | HiggsField |
| Video generation | HiggsField |
| Cinematic content | HiggsField |
| Ad creatives | HiggsField |
| Agentic coding | Codex |
| Deep research | Perplexity |
| Summarizing YouTube videos | NotebookLM |
| Real-time trends | Grok |
| Automations | Claude Code |
The 8 tools, what each one is actually for
🧠 1. Claude Opus 4.8 → deep thinking + strategy
My default for anything that needs a real brain. Strategy, positioning, scripts, copy, audit work, content critique. Opus 4.8 is the one I default to for anything I'd normally tap a smart friend on the shoulder for.
What I use it for in practice: drafting offer angles, pressure-testing decisions, writing carousels, refining scripts before I film.
Start here: claude.ai, pick Opus as the model.
🎨 2. Claude Design → design
Claude's dedicated design tool. I use it for design work where I want fast, on-brand output without context-switching to Canva or briefing a designer. Layouts, mock-ups, design iteration that lives next to the copy I'm writing.
What I use it for in practice: carousel designs, slide mock-ups, quick visual references for hires and editors.
🎬 3. HiggsField → image generation, video generation, cinematic content, ad creatives
HiggsField is my swiss army knife for visuals. Image generation, AI video generation, cinematic shot styles, and ad creative generation all live here. One tool, four jobs.
What I use it for in practice:
- Image generation when I need a specific scene or asset
- B-roll and AI video clips for reels
- Cinematic openers (the kind of shot that makes a viewer stop)
- Ad creative variations for paid testing
Start here: higgsfield.ai.
👁️ 4. Codex → agentic coding
OpenAI's agentic coding tool. I use it for builds where I want a coding agent that can read my files, write code, and run commands.
When to reach for it: anything where I'd otherwise be context-switching between a chat tab and an editor for hours.
🔍 5. Perplexity → deep research
Google replacement. I use Perplexity for anything where I need a real answer with sources, not 10 blue links and a 4-paragraph SEO intro. Fast, cited, actually useful.
What I use it for in practice: competitor research, fact-checking claims in scripts, finding up-to-date stats for carousels, looking up tool comparisons.
📓 6. NotebookLLM → summarizing YouTube videos
Google's source-grounded notebook. You upload sources (YouTube videos, PDFs, articles, transcripts) and it answers questions only from those sources. No hallucinations.
What I use it for in practice: summarising long-form YouTube videos so I can pull insights without watching the whole thing.
Start here: notebooklm.google.com.
📈 7. Grok → real-time trends
Grok is the fastest way I've found to get a read on what's trending right now. Plugged into X, surfaces what's actually moving in real time, which is exactly what you want when you're trying to ride a trend before it cools.
What I use it for in practice: spotting topics to ride for trending reels, catching news cycles early, sanity-checking what's actually breaking vs what feels like it's breaking.
⚙️ 8. Claude Code → automations
This is the engine room. Claude Code is the CLI / IDE version of Claude that can read your files, write code, and run commands. I use it to build the automations my business runs on: content workflows, scripts that edit my videos, custom skills that handle the boring repeatable work.
The difference vs claude.ai: Claude Code can actually take action on your computer. It's how you go from "chatting with AI" to "AI building things for you".
Start here: install with npm install -g @anthropic-ai/claude-code, then run claude in your terminal.
How I chain them (a real example)
Monday morning content sprint:
- Perplexity to research the week's AI news
- NotebookLM to digest the top 3 long-form YouTube videos in my niche
- Grok to scan what's trending in real time on X
- Claude Opus 4.8 to turn all of that into 5 script angles
- Claude Design to mock up the carousel slides for the supporting post
- HiggsField to generate B-roll, image assets, and cinematic openers for the reels
- Claude Code to run my automations: schedule posts, trigger the ManyChat lead capture flow, handle the boring repeatable bits