Seven AI Tools Worth Adding to Your Stack

Prompt (paste into Gemini):
"Generate 5 YouTube thumbnail variations for a video titled
'The 7 AI Tools Worth Paying For in 2026'.
Requirements:
- Bold text overlay: keep it to 5 words max
- High contrast background (dark with bright accent)
- Include a human face placeholder (shocked/curious expression)
- Each variation should test a different layout:
1. Text left, face right
2. Face center, text above
3. Split screen with tool logos
4. Minimalist with just text and gradient
5. Before/after style comparison
Style: clean, modern, YouTube-optimized for mobile.
Output at 1280x720."

Prompt:
"Analyze the top 20 performing YouTube videos in the
AI tools / productivity niche from the last 30 days.

For each video, extract:
- Title and channel name
- View count and publish date
- Thumbnail style (describe it)
- Hook used in the first 15 seconds
- Core content framework (listicle, tutorial, review, etc.)
Then identify:
1. The 3 most common title patterns
2. Topics with high views but low competition
3. Gaps: topics the audience wants that nobody has covered well
4. Recommended video ideas I could create this week
My niche: AI tools for business owners and creators.
My audience: 25-45, non-technical, revenue-focused."

Prompt:
"Create a set of 4 Instagram carousel images for a post
about '7 AI Tools That Replaced My Entire Team'.
Slide 1: Hook image - bold white text on dark gradient
Text: '7 AI Tools. $140/month. Zero employees.'
Slide 2: Clean infographic showing the 7 tool icons arranged
in a grid with names below each
Slide 3: Before/after comparison
Left: 'Old way - 5 tools, 3 freelancers, $3K/month'
Right: 'New way - 7 AI subs, $140/month'
Slide 4: CTA slide
Text: 'Full breakdown in bio. Free cheat sheet.'
Style: black background, white text, minimal.
Accent color: electric blue (#0066FF).
Output each slide at 1080x1080."

Prompt:
"Give me a real-time market brief as of right now.
Cover:
1. MACRO: Breaking news in the last 6 hours affecting US
equities, crypto, or global markets. Include source tweets.
2. CRYPTO: BTC and ETH price action + sentiment. Top 3
narratives on crypto Twitter right now. Whale movements
or exchange flows.
3. AI SECTOR: Major announcements from Anthropic, OpenAI,
Google, Meta, or xAI in the last 24 hours. Link tweets.
4. SENTIMENT: Overall market mood on X - bullish, bearish,
or uncertain? Back it up with posts from 50K+ accounts.
5. ACTION ITEMS: Based on all of the above, the 3 things
I should watch today.
Format: bullet points, concise, with tweet links."

Prompt (use DeepSearch mode):
"Deep research on $NVDA (NVIDIA) right now.
I want:
1. Latest earnings data and forward guidance
2. What Wall Street analysts are saying THIS WEEK
(find specific analyst tweets and reports)
3. Retail trader sentiment on X - bullish or bearish?
Top 5 most-engaged tweets about NVDA from 48 hours

4. Insider trading filings in the last 30 days
5. Supply chain signals - TSMC orders, chip demand,
datacenter buildout tweets that affect NVDA
6. Key risk factors the market might be underpricing
Give me a buy/hold/sell framework based on the data.
Not financial advice - just organize the information."

How to use (no prompt needed):
1. Find any tweet on X about a topic you care about
2. Click the Grok icon (star/sparkle) below the tweet
3. Grok automatically provides:
- Full context behind the tweet
- Related tweets and threads on the same topic
- Background on the person who posted it
- Counter-arguments or opposing views
- Links to primary sources
Pro tip: After Grok gives context, follow up with:
"Now compare this to what the top 5 accounts in [your niche]
are saying about the same topic. Who agrees? Who disagrees?
What is the consensus?"

- Open NotebookLM (notebooklm.google)
- Create a new notebook: "Viral Content Analysis"
- Click "Add Source" > paste YouTube URL of a viral video
- NotebookLM transcribes it automatically

"Break down this video into its structural components:
1. The hook (first 30 seconds) - what exactly did they say?
2. The promise - what did they tell the viewer they'd get?
3. Content framework - listicle, tutorial, review, story?
4. Retention tricks - pattern interrupts, open loops, cliffhangers
5. CTA - how and when did they ask for engagement?
6. Pacing - how long is each section?
Give me a template I can use to create my own version."

- Copy the analysis
- Paste into a Claude Project with your voice guide
- Ask Claude to write the full script in your voice

- Add up to 50 sources per notebook (PDFs, docs, URLs)
- Example: upload 5 research papers on a new topic

"Create 20 flashcards covering key concepts from all sources.
Each flashcard:
- Front: A clear question
- Back: A concise answer (2-3 sentences max)
- Source: Which document this came from
Focus on concepts I'd need to explain to someone else."

- Click "Audio Overview" in the Studio panel
- NotebookLM generates a podcast-style discussion
- Download and listen on the go
- Free: 3/day | Pro: 20+/day

Setup: Add 5-10 sources on a single topic
Prompt:
"Based on everything in this notebook:
1. What are the 3 most important findings that appear
across multiple sources? Cite which sources agree.
2. Where do the sources contradict each other?
List specific disagreements.
3. What questions are NOT answered by any of these sources?
4. If I had to brief someone in 2 minutes on this entire
topic, what would I say? Write that brief for me.
5. What are the strongest arguments on each side of the
central debate in these sources?"

Prompt (Perplexity Computer):
"Create an automated daily workflow that runs at 6:30am UTC.
The workflow should:
1. Scan the top 50 financial news sources for market-moving
events from the past 12 hours
2. Check crypto: BTC, ETH, SOL price action + on-chain events
3. Scan X for the top 10 most-engaged posts about AI, crypto,

and macro markets
4. Summarize into a 500-word brief with:
- 3 key headlines
- Market snapshot (indices + crypto)
- Sentiment assessment (1 sentence)
- 3 things to watch today
- Risk alerts (anything unusual)
Deliver to:
- Email: [your email]
- Slack: #morning-brief
- Discord: #market-updates
After each delivery, evaluate what I open and read.
Over time, prioritize content I engage with and deprioritize
what I skip."

Prompt (Perplexity Computer):
"Create a research report on AI infrastructure investments.
Phase 1 - Data Collection (use research models):
- Quarterly earnings from NVDA, AMD, MSFT, GOOG
- Total capex committed to AI infrastructure in 2025-2026
- Revenue growth rates for cloud AI services
- VC funding data for AI startups this quarter
Phase 2 - Analysis (use reasoning models):
- Compare growth rates across the top 5 AI infra companies
- Identify who is gaining or losing share
- Flag risks: overbuilding, demand gaps, margin pressure
Phase 3 - Output (use writing models):
- Format as a 2000-word executive brief
- Include a summary table comparing all companies
- Add 'Key Takeaways' with 5 bullet points
- End with 'What to Watch' for next quarter
Save as PDF and send to my email."

Prompt (Perplexity Computer):
"Build a continuous monitoring workflow for real estate
opportunities in Dubai Marina and JBR.
Parameters:
- Type: 1-2 bedroom apartments
- Price: AED 800K - 1.5M
- Yield: minimum 7% gross rental
- Ready to move in (no off-plan)
Workflow:
1. Scan Property Finder, Bayut, Dubizzle daily
2. Filter listings matching my criteria
3. Calculate estimated rental yield from market data
4. Score each property 1-10 (location, price/sqft, view, floor)
5. Weekly digest: top 5 finds
6. Immediate alert if anything scores 8+
Include listing link, photos, and analysis for each.
Track price changes on listings I have seen before."

MindOS Agent Setup:
Agent Name: "Competitor Intel Agent"
Agent Role: YouTube competitive analyst
Instructions:

"You are my competitive analysis agent for YouTube.
Every day at 8am:
1. Visit these 5 competitor channels: [list URLs]
2. Find their most recent 3 videos each
3. Record: title, views, publish date, like count
4. Identify which video outperformed relative to
their average (views vs. subscriber count)
5. Why did it outperform? Topic, hook, thumbnail?
Weekly summary (every Monday):
- Which competitor grew the most this week?
- Top 3 performing videos across all competitors
- Topics with high engagement that I have NOT covered
- 3 video ideas based on gaps I can fill
Deliver via email. Keep it under 400 words."

MindOS Agent Setup:
Agent Name: "Deal Finder"
Agent Role: Property and investment scout
Instructions:
"You are my deal research agent.
Scan daily: Rightmove, Zoopla, OnTheMarket (UK)
Property Finder, Bayut (Dubai)
Criteria:
- Budget: $200K-500K
- Type: apartments, 1-2 bed
- Yield: minimum 6% gross
- Location: [your target areas]
For each match:
1. Extract price, size, location, photos
2. Estimate rental yield from comparable listings
3. Check if the price dropped recently
4. Score 1-10

Send me only listings scoring 7+.
Include the link and a 2-sentence summary of why
it is worth looking at."

MindOS Agent Setup:
Agent Name: "Content Repurposer"
Agent Role: Cross-platform content distribution
Instructions:
"You are my content repurposing agent.
When I publish a new YouTube video:
1. Go to my channel, find the latest video
2. Extract the title, description, and transcript
3. Create a Twitter/X thread (7-10 tweets) with a hook
4. Create a LinkedIn post (300 words, professional)
5. Create an Instagram carousel outline (8 slides)
6. Save all drafts to a Google Doc titled:
'[Video Title] - Repurposed Content'
Rules:
- Each platform should feel native, not copy-pasted
- Twitter: punchy, conversational, numbered
- LinkedIn: thought-leadership, first-person
- Instagram: visual-first, one idea per slide
- Always end with a CTA to the video"

Prompt:
"You are my business strategy advisor. Context:
- Revenue: $1.8M/year, growing 15% MoM
- Business: AI media company, 30 employees
- Channels: YouTube (500K subs), X (1.2M followers),
newsletter (200K subscribers)
- Revenue: sponsorships (60%), courses (25%), consulting (15%)
- Goal: $5M/year within 12 months
Questions:
1. Highest-leverage growth move in the next 90 days?
2. Which revenue stream scales without adding headcount?
3. Where am I most vulnerable? What could break?
4. If you were my COO, what would you do in week one?
5. 90-day sprint plan with 3 measurable milestones.
Be direct. Challenge my assumptions. Don't sugarcoat."

Voice prompt examples (say out loud to ChatGPT):
Meeting prep:
"I have a meeting in 20 minutes with a potential sponsor.
They run a SaaS company in AI. They want to sponsor my
YouTube channel. What 3 questions should I ask? What
should my minimum deal be for 500K subscribers? Red flags?"
Idea capture:
"I just had an idea for a video. The concept: showing the
exact dollar amount I spend on AI tools per month and the
ROI of each one. Hook: 'I spend $347/month on AI. Here is
what I get back.' Does this work? How would you structure
it for maximum retention?"
Quick brainstorm:
"10 YouTube video titles about AI tools that would get
clicks. Curiosity-driven. Include numbers.
Avoid clickbait that doesn't deliver."

Codex cloud prompt:
"In my repo (link to GitHub), create:
1. New API endpoint /api/analytics returning JSON summary
of user engagement metrics
2. Pull from existing PostgreSQL database
3. Include: DAU, avg session length, top 5 pages, conversion
4. Add error handling and rate limiting
5. Write tests for the new endpoint
6. Create a PR with clear description
Use existing code style. Do not modify existing endpoints."
Codex CLI (local):
$ codex "Refactor the auth module to use JWT tokens instead
of session cookies. Update all routes and middleware.
Run test suite after changes and fix any failures."

# Start Claude Code
cd ~/my-project && claude
Prompt:
"Build me a personal finance dashboard.

Requirements:
- Read transactions from a CSV I will provide
- Categorize each transaction (food, transport, subs, etc.)
- Store in local SQLite database
- Web dashboard on localhost:3000 showing:
- Monthly spending by category (bar chart)
- Spending trend over 6 months (line chart)
- Top 10 merchants by total spend
- Recurring subscriptions detected automatically
- Alerts if any category exceeds average by 20%
Tech: Python backend, HTML/JS frontend, Chart.js.
Black and white design. No frameworks."

Project folder:
/YouTube-Content/
INSTRUCTIONS.md
MEMORY.md
CONTEXT-brand-voice.md
CONTEXT-past-scripts.md
/outputs/
INSTRUCTIONS.md:
# INSTRUCTIONS
## Who You Are
You are my YouTube content strategist and scriptwriter.
## What You Do
- Generate video ideas based on trending topics
- Write full video scripts in my voice
- Review and score thumbnail concepts
- Create titles and descriptions optimized for CTR
## Rules
- Match my tone: direct, practical, no fluff
- Never invent statistics - flag anything needing a source

- Every script: hook, promise, body, CTA
- If unclear, ask ONE question before starting
- Log new preferences to MEMORY.md
## What Good Output Looks Like
- Scripts: 2000-3000 words, conversational, with timestamps
- Titles: 8-12 words, curiosity-driven, include a number
- Thumbnails: score 1-10 with specific feedback
## Memory
Start: read MEMORY.md
End: update MEMORY.md with new learnings

Artifact prompt:
"Create an interactive HTML dashboard showing my content
performance across platforms.
Include:
- YouTube: views, subs gained, top 5 videos this month
- X/Twitter: impressions, engagement rate, follower growth
- Newsletter: open rate, click rate, subscriber count
Bloomberg terminal aesthetic. Dark background. Green accent.
Monospace for numbers. Date selector for time range.
Sparkline charts for trends."
Cowork prompt (from phone via Telegram):
"Open my Google Sheets revenue tracker. Add a new row:
$4,200 sponsorship from [Company], invoice sent,
payment due 30 days. Update the monthly total formula."
Dispatch auto-routing:
- "Research this company" -> triggers web research
- "Write a draft about this" -> triggers writing project
- "Build a tool for this" -> triggers Claude Code