# 1. Jan 2026_AI Resource Hub

## Header
AI for Marketing Resource Hub
**AI is transforming marketing, and I'm learning alongside you.**
I built this to document my learning by collecting the best resources I'm actually using and adding my personal perspective on what makes each one valuable for marketers. <br><br>I update this regularly, as I find and test what works. Join me on <a href="https://www.linkedin.com/in/yzsun/">LinkedIn</a> to swap ideas, suggest additions, or just say hi.

## Navigation
Jump to:
| Label | Link URL |
| --- | --- |
| Videos | #videos |
| Research | #reports |
| Podcasts | #podcasts |
| Newsletters | #newsletters |
| Courses | #courses |
| AI Applications | #applications |

## Videos
Featured Videos
These are the AI and marketing videos I keep coming back to. I selected them for their range and depth, from AI builders to practitioners using AI in the field. They're long enough to matter and rich enough to revisit. 
:icon-youtube:
https://www.youtube.com/embed/
No videos available for this month yet.
Video Placeholder
| Tab Name | Title | Description |
| --- | --- | --- |
| May 2026 | The Agentic Marketing Era Begins  | May goes deeper into agentic AI. From defining what AI agents actually are to seeing how they operate across the marketing funnel, the question this month isn't whether agentic AI will change marketing. It's whether we'll be ready when it does. |
| April 2026 | AI Search Visibility: Foundation, Insight, and Action | April's Reddit research identified AI search visibility as the most discussed and upvoted topic among marketers. These videos explain what AI search is, why it matters, how experts approach it, and how it actually works. |
| March 2026 | Specialized Tools, Smarter Usage | March digs into specialized applications: AI search, document analysis, and visual creation, then steps back to ask the bigger questions: how to boost AI productivity and how to think more strategically about AI tools? |
| February 2026 | AI Tools Meet Marketing Strategy | February focuses on application. I picked these videos to showcase how top marketers leverage leading LLM tools, alongside three deep dives into the strategies and trends shaping digital marketing in 2026. |
| January 2026 | AI &amp; Marketing: The Full Landscape  | Real understanding comes from range. January's videos bring together marketing legends, AI founders, top agencies, and academic voices to create a fuller view of AI in Marketing. |
| Tab Name | Title | Description | YouTube Video ID | Video URL |
| --- | --- | --- | --- | --- |
| January 2026 | Dr Diane Hamilton Interviews Seth Godin | Seth's advice hit home: Don't wait for permission to be curious. Use AI for minor tasks and step into the innovator role. | CtU064MfcuU | https://www.youtube.com/watch?v=CtU064MfcuU |
| January 2026 | How to Go From Beginner to Advanced with AI | A clear 4-level AI mastery that makes sense for non-technical people. Check if you've hit level 2 yet. | BZuXGSm1lN8 | https://www.youtube.com/watch?v=BZuXGSm1lN8 |
| January 2026 | How I Use LLMs | <span style="color: rgb(19, 19, 19);">Don't miss Andrej Karpathy's intro videos! In this one, he shares his actual use cases across different LLMs.</span> | EWvNQjAaOHw | https://www.youtube.com/watch?v=EWvNQjAaOHw |
| January 2026 | The New SEO Playbook for AI Search | <span style="color: rgb(19, 19, 19);">In under 9 minutes, I learned 5 core areas that give websites the best chance of ranking in AI search with GEO. </span> | gReszNnykpg | https://www.youtube.com/watch?v=gReszNnykpg |
| January 2026 | How AI Could Change the Advertising Business | <span style="color: rgb(19, 19, 19);">See how WPP uses AI in the agency world: synthetic focus groups, massive content ideation, and brand engagement prediction.</span> | a40YkQDSIrk | https://www.youtube.com/watch?v=a40YkQDSIrk |
| January 2026 | Leveraging AI for Marketing | <span style="color: rgb(19, 19, 19);">Professor Gupta explains </span>AI's impact through the 4 Ps. Key insight: AI enables hyper-personalization, but may eventually drive up the cost of consumer attention. | YQvBScJGoEI | https://www.youtube.com/watch?v=YQvBScJGoEI |
| February 2026 | How Anthropic uses Claude in Marketing | Advice from Anthropic growth marketer: identify repetitive tasks, start small, and focus on explaining the challenge and desired outcome. | Jp83_JMK74o | https://www.youtube.com/watch?v=Jp83_JMK74o |
| February 2026 | 9 ChatGPT Use Cases from the Top 1% of Marketers | Great use cases across product, content, and growth. Highlight: an agent that monitors community conversations and generates LLM-optimized landing pages at scale. | ZCWJkWHOXLU | https://www.youtube.com/watch?v=ZCWJkWHOXLU |
| February 2026 | Google's New Gemini is Revolutionizing Marketing | A practical walkthrough of Gemini marketing use cases, from research synthesis in NotebookLM, to data analysis in Gemini, and to rapid content repurposing with Gems. | SgK9lgxWPzs | https://www.youtube.com/watch?v=SgK9lgxWPzs |
| February 2026 | AI-ready Marketing: The Next Shift in Digital Marketing Strategy | Being AI-ready starts with the basics: high-quality first-party data, integrated platforms, and incrementality testing to measure the actual uplift. | vw9dqka3oDg | https://www.youtube.com/watch?v=vw9dqka3oDg |
| February 2026 | 8 Trends I’m Betting My Entire Marketing Strategy On in 2026 | Two trends stood out to me: On-platform lead gen is gaining ground across social. Schema markup (structured data) is essential for browser-level AI visibility. | hXPALnu3Y6I | https://www.youtube.com/watch?v=hXPALnu3Y6I |
| February 2026 | How to Build a Page AI Will Love (AEO Checklist) | Google indexes the whole page, but LLMs read in chunks. Tips: put the answer first, use clear structure, and make every passage work in isolation. | Yy9ksx2Lygc | https://www.youtube.com/watch?v=Yy9ksx2Lygc |
| March 2026 | Perplexity | It’s a real-time research analyst cross-referencing live web with internal data to ensure marketing strategy is as evidence-based as it is creative. | GJgKC5yT7To | https://www.youtube.com/watch?v=GJgKC5yT7To |
| March 2026 | NotebookLM | It turns mountains of information into clear marketing insights by grounding AI in specific project data and asking strategic questions. | 6M3PesCgTz0 | https://www.youtube.com/watch?v=6M3PesCgTz0 |
| March 2026 | Midjourney | It's an engine for iterating on ideas that explores visual possibilities and masters the art of prompt engineering to bring any creative assets to life. | LNbDHGZR1Zk | https://www.youtube.com/watch?v=LNbDHGZR1Zk |
| March 2026 | AI Productivity Playbook | Mastering AI isn't about better prompts; it's about better coaching. Jeremy Utley shows how to treat AI as a strategic partner. Highlight? Using roleplaying to prepare for a difficult conversation. | yMOmmnjy3sE | https://www.youtube.com/watch?v=yMOmmnjy3sE |
| March 2026 | Become an AI-Native | AI-first task planning: breaking complex projects into small tasks to decide what stays manual and what gets handed to AI before starting. Highlight? Selecting the right AI for the right kind of work. | E7YiKBeOneo | https://www.youtube.com/watch?v=E7YiKBeOneo |
| March 2026 | Strategy before Tactics  | Move from stacking tools to using AI as a strategic partner to solve critical business friction points. Highlight? Real-time audience question mining that uncovers actionable customer pain points in minutes. | jG_5jLmvh88 | https://www.youtube.com/watch?v=jG_5jLmvh88 |
| April 2026 | AI SEO Foundation | Traditional SEO isn't dead! It's the foundation AI search crawls. This framework teaches intent-driven keyword evaluation, then shows how to build topical authority cluster by cluster. | 7DRO4rEIHDk | https://www.youtube.com/watch?v=7DRO4rEIHDk |
| April 2026 | Ultimate AEO Guide | Ethan Smith explains why citation density beats single rankings in AI search. Focus on: answering hyper-specific long-tail questions, building citation presence across Reddit and YouTube, and tracking share of voice.  | iT7kq-R3Gjc | https://www.youtube.com/watch?v=iT7kq-R3Gjc |
| April 2026 | GEO Expert Playbook | Top tactics: analyze sales transcripts to uncover actual customer questions, create niche comparison pages, and build third-party credibility through Reddit mentions and industry listicles. | RwKKLnyXCig | https://www.youtube.com/watch?v=RwKKLnyXCig |
| April 2026 | Ranked #1 in ChatGPT | Real B2B SaaS results: 5x trial growth. CITABLE framework: clear BLUF structure, intent-driven pillar content, third-party validation, answer grounding with sources, block structure for RAG extraction, latest timestamps, and entity mapping. | eSBmFv7jb9Q | https://www.youtube.com/watch?v=eSBmFv7jb9Q |
| April 2026 | Rank in AI Search | Earn brand mentions on industry lists and reviews, create AI-preferred content (how-to guides, comparisons), use BLUF structure for atomic extraction, and dominate Reddit and YouTube where AI models cite most. | DnRCo2vvKvY | https://www.youtube.com/watch?v=DnRCo2vvKvY |
| April 2026 | New Research Data | Financial services see 2.9x AI search growth driven by education-first discovery. Users spend 63% of sessions learning before transacting. Winners focus on contextual educational content, structured data, and entity linking. | __ApTxTzkdc | https://www.youtube.com/watch?v=__ApTxTzkdc |
| May 2026 | Agentic AI Explained | McKinsey breaks down what agentic AI is, how it differs from traditional automation, and what enterprise deployment really requires. The standout concept: agentic mesh that prevents teams from building isolated agent silos as they scale. | DsRNp3cmJm0 | https://www.youtube.com/watch?v=DsRNp3cmJm0 |
| May 2026 | Agentic AI for Marketers | Agentic AI moves beyond automation by reasoning, evaluating trade-offs, and delivering decision-ready intelligence. A practical framework that bridges the gap between data and execution by answering what changed, why it matters, and what to do next. | dFBL6dFri7U | https://www.youtube.com/watch?v=dFBL6dFri7U |
| May 2026 | Build and Measure AI Agents | A practical blueprint for moving from agentic AI theory to execution. Covers how to document and evaluate workflows, how to match tasks to the right precision level, and how to track agent performance across efficiency, quality, and business impact metrics. | ibFJ--CH3cQ | https://www.youtube.com/watch?v=ibFJ--CH3cQ |
| May 2026 | Three Domains of AI Agents in Marketing | Scott Brinker brings the same landscape-mapping lens to agentic AI, introducing a three-archetype model: agents for marketers, agents for customers, and agents of customers. The last one is reshaping how brands get discovered, evaluated, and chosen. | q6lOCmawkDg | https://www.youtube.com/watch?v=q6lOCmawkDg |
| May 2026 | Agentic AI in Marketing Measurement | GrowthLoop's CEO walks through how AI agents synthesize first-party data, automate audience building, and shift marketing measurement from vanity metrics to provable incrementality, positioning marketing as a verified growth engine. | prB9G9NgWDU | https://www.youtube.com/watch?v=prB9G9NgWDU |
| May 2026 | Google Marketing Live 2026 | The agentic era arrived and it touched everything. Search is now conversational. Ads are now interactive. YouTube is now shoppable end to end. And Ask Advisor brings a single AI partner with shared memory across Google's entire marketing platform. | HjMHj0uLI1k | https://www.youtube.com/watch?v=HjMHj0uLI1k |

## Reports
Research & Insights
I rely on top consulting firms and leading business publications for hard data, industry benchmarks, and credible trends. These selections cover AI's evolving impact on marketing from strategy to implementation.
:icon-file-text:
| Report Title | Source/Author | Summary | Link | Is New |
| --- | --- | --- | --- | --- |
| Reinventing marketing workflows with agentic AI | McKinsey | Agentic AI could power up to 60% of marketing tasks yet fewer than 10% of organizations have captured value beyond isolated pilots. This piece walks through a five-step framework for designing the hybrid human-agent workflows. | https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights/reinventing-marketing-workflows-with-agentic-ai | Yes |
| A new generation of consumers, a new era of AI | Pwc | Agentic commerce and Gen Alpha will redefine the retail funnel by 2026, requiring brands to pivot toward machine-readable data and authentic, human-centric engagement to maintain discoverability in an AI-governed marketplace. | https://www.pwc.com/us/en/industries/consumer-markets/library/sxsw-event-recap.html | No |
| The State of AI in the Enterprise – 2026 AI Report | Deloitte | Workplace AI access rose 50% in 2025, and the share of companies with 40%+ of AI projects in production is set to double within six months, accelerating scale in front-office functions like marketing. | https://www.deloitte.com/us/en/what-we-do/capabilities/applied-artificial-intelligence/content/state-of-ai-in-the-enterprise.html | No |

## Podcasts
Podcasts
I use podcasts to continue learning and keep up with new ideas while on the go. These four deliver real insights from industry leaders who are building and implementing with AI.<br>
:icon-mic:
| Podcast Name | Description | Link |
| --- | --- | --- |
| Marketing Against the Grain | My one-stop show for the latest news, deep insights, marketing leader interviews, and product demos. | https://www.youtube.com/playlist?list=PLJ7YXFIg84xB0Na0vQKe9uBtnNXV7kaT7 |
| Decoding AI for Marketing | In-depth interviews with leading marketing and AI insiders who bridge technical complexity with real-world examples. | https://www.youtube.com/playlist?list=PLCitlSxQvKIiXQx5QxhI7dX27a2jkI4un |
| Leveling Up | I turn to Eric Siu for applicable AI tactics. Running an agency gives him the advantage of testing constantly.   | https://www.youtube.com/playlist?list=PLWSSOx1A_2vWDWStRB2Bf3oMHLOceEHX- |
| AI &amp; I  | My favorite for learning how top founders and creators actually integrate AI into their work and lives. | https://www.youtube.com/playlist?list=PLuMcoKK9mKgHtW_o9h5sGO2vXrffKHwJL |

## Newsletters
Newsletters
**Staying sharp on marketing and AI does not mean drowning in content.**<br><br>The top two keep me grounded in marketing fundamentals and industry shifts, while the bottom two help me track what is changing in AI without the hype.
:icon-mail:
| Newsletter Name | Description | Link |
| --- | --- | --- |
| <span style="color: rgb(31, 31, 31);">TLDR Marketing</span> | Perfect mix of marketing and AI. Miscellaneous is where I found the best surprises. | https://tldr.tech/marketing |
| Marketing Brew | My go-to for tracking how brands are adapting to platform changes and new ad tech in real time. | https://www.marketingbrew.com/ |
| Superhuman AI | <span style="color: rgb(34, 34, 34);">Fast AI news digest that keeps me current on tool releases and model updates without eating up my day.</span> | https://www.superhuman.ai/ |
| The Batch | <span style="color: rgb(50, 50, 50);">Weekly analysis that goes beyond AI news to explain how things work and why they matter. Plus, Andrew Ng's take.</span> | https://www.deeplearning.ai/the-batch/ |

## Courses
AI Courses
For structured learning, I look for courses from leading organizations. These three are my starting points for understanding AI fundamentals, GenAI, and AI applications in marketing. 
:icon-library-big:
| Report Title | Source/Author | Summary | Link |
| --- | --- | --- | --- |
| AI Essentials<br> | <span style="color: rgb(31, 31, 31);">Google</span> | · Beginner level<br>· 4 hours to complete<br><br>My entry point into AI learning. It teaches fundamentals with a practical prompting framework and quality resources to stay ahead of AI trends. | https://www.coursera.org/google-certificates/ai-essentials-google |
| Generative AI for everyone | <span style="color: rgb(31, 31, 31);">DeepLearning.AI</span> | · Beginner level<br>· 3 hours to complete<br><br>Deep dive into GenAI. The job analysis framework (breaking down tasks by AI potential) completely shifted how I think about marketing automation.  | https://www.deeplearning.ai/courses/generative-ai-for-everyone/ |
| AI for Marketing | HubSpot | · Beginner level<br>· <span style="color: rgb(33, 51, 67);">3 hours</span> to complete<br><br>The most complete marketing-focused AI course I've found so far. It includes real use cases and brings in marketing leaders to share what's actually working.<br>&nbsp; | https://academy.hubspot.com/courses/AI-for-Marketers |

## Applications
AI Applications
A collection of battle-tested prompts and Claude Skills from my marketing case studies. These have delivered real results. Give them a try and see what they can do for your work.
:icon-sparkles:
Show more
Copy to Clipboard
Copied!
| Label |
| --- |
| All |
| Prompt |
| Claude Skills |
| Category | Badge Text | Title | Content/Code |
| --- | --- | --- | --- |
| Prompt | PROMPT | Market Research (Gemini Deep Research) | <p>OBJECTIVE: Conduct a comprehensive market research on the AI-powered smart glasses market to inform Warby Parker's product launch strategy in late 2026.</p><p><br>RESEARCH PARAMETERS:</p><p>Geographic focus: North America (US and Canada)</p><p>Product category: AI-powered smart glasses</p><p>Time horizon: 2026-2030<br><br>AREAS OF FOCUS FOR THE RESEARCH:</p><p><br>FOCUS 1: MARKET LANDSCAPE Purpose: Understand the AI glasses market opportunity and dynamics</p><p>Research:</p><p>Market size and opportunity</p><p>Market growth trajectory</p><p>Market structure</p><p>Category maturity</p><p>Adoption barriers and enablers</p><p><br>FOCUS 2: CUSTOMER ANALYSIS Purpose: Understand who buys AI glasses and why</p><p>Research:</p><p>Customer segmentation</p><p>Jobs-to-be-Done</p><p>Customer needs and pain points</p><p>Purchase barriers</p><p>Willingness to pay and price sensitivity</p><p><br>FOCUS 3: COMPETITOR ANALYSIS Purpose: Map the competitive battlefield</p><p>Research:</p><p>Competitive landscape</p><p>Competitive positioning</p><p>Competitor strategies</p><p>Competitor performance and success metrics</p><p>Pricing landscape</p><p><br>FOCUS 4: AI GLASSES PRODUCT LINE EXPANSION (SWOT) Purpose: Assess Warby Parker's position entering this product category</p><p>Research:</p><p>Strengths of launching AI glasses</p><p>Weaknesses of launching AI glasses</p><p>Opportunities for this product line</p><p>Threats to this product line</p><p><br>FOCUS 5: GO-TO-MARKET CONSIDERATIONS Purpose: Synthesize insights into strategic implications</p><p>Research:</p><p>Launch readiness factors</p><p>Media and influencer landscape</p><p>Key strategic insights</p><p>Strategic implications</p><p><br>REPORT STRUCTURE: Present findings as a comprehensive market research report organized by the 5 focus areas above. Provide data-driven insights with specific metrics and examples where available.</p> |
| Prompt | Prompt | Campaign Ideation (ChatGPT) | <p>PRODUCT CONTEXT: Warby Parker is launching AI-powered smart glasses in late 2026, developed with Google and Samsung. These are prescription glasses with built-in AI capabilities (voice assistance, real-time information, hands-free features) designed for all-day wear.</p><p><br>Key product features:</p><p>Prescription lenses with AI integration</p><p>Google Gemini AI assistant</p><p>Lightweight, designed for all-day comfort (under 50 grams)</p><p>Available in Warby Parker retail stores for try-on and prescription fitting</p><p>Positioned as "accessible luxury"</p><p><br>TARGET AUDIENCE: <br>North American professionals and tech-forward consumers who already wear glasses and value both design and functionality. Emerging segment: "Digital Minimalists" seeking to reduce smartphone dependency.</p><p><br>KEY MARKET INSIGHTS: <br>-The "Prescription Moat": Smart glasses require optical retail infrastructure for prescription fitting—a barrier tech giants like Meta and Google can't easily replicate. This positions Warby Parker's 323+ retail stores as a unique competitive advantage.</p><p>-Meta's 82% market share came from strategic restraint: Meta Ray-Ban succeeded by avoiding complex features (no heads-up displays), prioritizing style and simplicity, and staying under the $400 psychological price barrier to solve the "looking like a prototype" stigma.</p><p>-2026 is the mainstream inflection point: Industry projections show 2026 as the "breakout year" with global sales exceeding 10 million units, driven by lightweight designs (under 50 grams) and all-day battery enabled by on-device AI.</p><p>-The emerging "Digital Minimalist" segment: Growth is coming from users who want to reduce smartphone friction, not add more screens. They're hiring smart glasses for "glanceable" AI information to stay connected without being tethered to their phones.</p><p>-Google is de-risking Warby Parker's entry: Google's $150M commitment (development costs + optional investment) allows Warby Parker to innovate with significantly lower financial risk than a traditional hardware startup.</p><p><br>MARKETING OBJECTIVE: <br>Generate awareness and excitement for the product launch while differentiating from tech-first competitors like Meta Ray-Ban.</p><p><br>TASK: <br>Generate 5 distinct campaign concepts for the launch. Each concept should include:</p><p>A unique strategic angle or positioning</p><p>A campaign tagline or headline</p><p>A brief description of the core message and creative direction</p><p>The type of customer this angle would resonate with most</p><p><br>CREATIVE REQUIREMENTS:</p><p>Make each angle meaningfully different (not just slight variations)</p><p>Mix emotional and rational appeals</p><p>Include at least one contrarian or unexpected angle</p><p>Vary tone: aspirational, practical, identity-driven, etc.</p><p>Consider Warby Parker's brand heritage (accessible luxury, design-forward, challenger brand)</p><p>Use the key market insights to inform differentiation from Meta</p><p><br>AVOID:</p><p>Generic "future of technology" messaging</p><p>Overly technical feature lists</p><p>Copying Meta's approach (simple, social-first positioning)</p><p>Ignoring the Digital Minimalist insight or prescription advantage</p> |
| Claude Skills | Claude Skills | AI Search Tracking (Claude) | <p>Automates Google AI Mode (google.com/ai) search result tracking and competitive intelligence gathering. Use this skill whenever a user wants to: track what Google AI Mode says about a topic, monitor which brands or domains get cited by Google AI answers, capture AI-generated responses for competitive research, or analyze citation patterns across multiple search queries. Trigger on phrases like "track AI search results", "what does Google AI say about", "capture Google AI Mode responses", "monitor AI citations", "AI search tracker", or any request to search Google AI Mode and save/analyze the results. This skill handles the full workflow: collecting queries from the user (if not already provided), running each query in a live Chrome browser, extracting the AI response and citations, analyzing brand frequency, and producing a Word report and an Excel tracker — all automatically.</p><p>---</p><p># Google AI Mode Search Tracking</p><p>## Overview</p><p>This skill automates a repeatable research workflow: run one or more queries in Google AI Mode, capture the full AI-generated response and top cited sources for each, then produce two polished output files —a Word document (narrative report with a brand frequency summary) and an Excel spreadsheet (structured citation log). It also flags which brands and domains appear most often across all queries, which is useful for competitive intelligence and SEO research.</p><p>---</p><p>## Step 1 — Collect Queries</p><p>Before doing anything else, check whether the user has already provided their search queries.</p><p>**If queries were provided in the user's message** (e.g., listed as bullet points, a numbered list, or inline in the request), extract them directly and confirm with the user before proceeding.</p><p>**If no queries were provided**, ask for them explicitly. Use the `AskUserQuestion` tool with a text input prompt — don't just type a question in your response, because the input box makes it easy for the user to paste a list. A good prompt:</p><p>&gt; "Please provide your search queries — paste them as a list, one per line. Example:\n</p><p>&gt; smart glasses for everyday use\n</p><p>&gt; best smart glasses with AI features"</p><p>Do not proceed to browser steps until you have at least one confirmed query.</p><p>---</p><p>## Step 2 — Set Up Browser</p><p>1. Use `tabs_context_mcp` to check whether a Chrome tab is available. If not, create one with</p><p>   `tabs_create_mcp`.</p><p>2. Navigate to `https://google.com/ai`. Wait for the "Ask anything" input to appear.</p><p>3. If a modal or popup appears (e.g., "Supercharge Search with Personal Intelligence"), close it by clicking the X button or "Not now" before proceeding.</p><p>4. Confirm the AI Mode interface is loaded (you should see the "Ask anything" text box and the "AI Mode" label in the top navigation).</p><p>---</p><p>## Step 3 — Run Each Query</p><p>For **each query** in the list:</p><p>### 3a. Submit the query</p><p>- Click the "Ask anything" text box and type the query.</p><p>- Press Enter and wait for the AI response to fully load (typically 5–8 seconds). Take a screenshot to verify the response has rendered before extracting data.</p><p>### 3b. Extract the AI response</p><p>Use `read_page` with `filter: "all"` and `depth: 8` to capture the full accessibility tree. This reliably captures the AI-generated text, inline citations (hyperlinked brand names), and the cited source list — even content that isn't visible in the viewport.</p><p>From the accessibility tree, extract:</p><p>- The full AI-generated response text (all paragraphs, headings, bullet points, and tables)</p><p>- Inline citation labels (e.g., "Wired +1", "Meta +5") that appear within the response</p><p>### 3c. Extract cited sources</p><p>Look for a "Show all related links" button in the accessibility tree and click it to expand the full citation list. Then re-read the dialog that opens (use `read_page` with `ref_id` pointing to the dialog element).</p><p>From the citation list, extract for each source:</p><p>- **Page title** (the link text, e.g., "The Best Smart Glasses to Augment Your Reality - WIRED")</p><p>- **Source brand** (the publisher label below the title, e.g., "WIRED")</p><p>- **Full URL** (the `href` attribute of the link)</p><p>- **Domain** (extract from the URL, e.g., `wired.com`)</p><p>Capture the **top 5 cited sources** (first 5 in the list). If fewer than 5 exist, capture all.</p><p>### 3d. Move to the next query</p><p>After capturing all data for this query, click the "Ask anything" text box at the bottom of the page and type the next query. Do not navigate away — the AI Mode interface allows follow-up queries in the same session.</p><p>---</p><p>## Step 4 — Analyze Brand Frequency</p><p>Once all queries are complete, count how many times each **domain** appears across all citation lists (combined). This tells you which sources Google AI cites most often.</p><p>Build a ranked table:</p><p>- **Brand / Domain** (e.g., "WIRED(wired.com)")</p><p>- **Total Citations** (count across all queries)</p><p>- **Queries Appeared In** (which queries cited this domain)</p><p>Sort by total citations descending. Highlight domains with 3+ citations as high-frequency.</p><p>---</p><p>## Step 5 — Generate Output Files</p><p>Read the skill files for creating Word and Excel documents before writing any code:</p><p>- `/sessions/confident-gallant-brown/mnt/.skills/skills/docx/SKILL.md` for Word document creation</p><p>- `/sessions/confident-gallant-brown/mnt/.skills/skills/xlsx/SKILL.md` for Excel creation</p><p>### Word Document (`.docx`)</p><p>Use the `docx` npm package (always available — `npm install -g docx` if needed). Structure:</p><p>```</p><p>Cover / Title Page</p><p>  - Title: "Google AI Mode Search Results Tracker"</p><p>  - Subtitle: topic or date range</p><p>  - Date Tracked, number of queries run</p><p>One section per query (use Heading 1 for query text):</p><p>  - "AI-Generated Response" (Heading 2)</p><p>    → Full response text, preserving headings and bullet points</p><p>  - "Top Cited Sources" (Heading 2)</p><p>    → Table: Page Title \| Source Brand \| URL</p><p>Summary Section (Heading 1: "Brand &amp; Domain Frequency Analysis")</p><p>  - Intro paragraph</p><p>  - Frequency table: Brand/Domain \| Total Citations \| Queries Appeared In</p><p>  - Key Takeaways bullets (one per top brand, explaining what their citation frequency signals)</p><p>Footer: page numbers + "Internal Use Only"</p><p>Header: "Google AI Mode Search Tracker \| Date"</p><p>```</p><p>Use professional styling: navy/blue heading colors, alternating row shading on tables, Arial font.</p><p>Always set US Letter page size (12240 × 15840 DXA) with 1-inch margins.</p><p>### Excel Spreadsheet (`.xlsx`)</p><p>Use `openpyxl`. Create two sheets:</p><p>**Sheet 1 — "AI Search Results"**</p><p>Columns: `Query \| Cited Page Title \| Cited Page URL \| Cited Domain \| Source / Brand \| Date Tracked`</p><p>- One row per cited source per query (so 5 rows per query if 5 citations were captured)</p><p>- Freeze the header row (row 3 after title rows)</p><p>- Format URLs with blue underline font</p><p>- Alternate row shading for readability</p><p>- Auto-fit column widths</p><p>**Sheet 2 — "Domain Frequency Analysis"**</p><p>Columns: `Brand / Domain \| Domain \| Total Citations \| Queries Appeared In`</p><p>- Sorted by Total Citations descending</p><p>- Highlight cells in the "Total Citations" column with 3+ citations in red bold text</p><p>- Freeze header row</p><p>After generating both files, save them to `/sessions/confident-gallant-brown/mnt/Claude CoWork/` and present them to the user with `present_files`.</p><p>---</p><p>## Output File Naming</p><p>Use this naming convention so files stay organized when the skill is run multiple times:</p><p>```</p><p>Google_AI_Mode_Tracker_&lt;YYYY-MM-DD&gt;.docx</p><p>Google_AI_Mode_Tracker_&lt;YYYY-MM-DD&gt;.xlsx</p><p>```</p><p>---</p><p>## Error Handling</p><p>- **Chrome not connected**: Inform the user that Claude in Chrome must be active. Direct them to open Chrome, activate the Claude extension, and retry.</p><p>- **Google AI Mode not loading**: Try navigating directly to `https://google.com/search?udm=50`.</p><p>- **Response still loading**: If the accessibility tree shows "Thinking" or "Searching" states, wait 3 more seconds and re-read.</p><p>- **Popup blocking input**: Always check for and dismiss modals before typing queries. Use the</p><p>  X or "Not now" button.</p><p>- **Fewer than 5 citations**: Normal — capture all that are available and note the count.</p><p>---</p><p>## Tips for Accurate Extraction</p><p>- Google AI Mode renders citations inside complex nested accessibility trees with shadow DOM.</p><p>  The `read_page` tool is more reliable than `get_page_text` or JavaScript for extraction.</p><p>- The "Show all related links" button (or dialog) contains the full citation list. Always expand</p><p>  it before extracting — the 3-citation preview visible on the page is not the complete list.</p><p>- If `read_page` output is too large, use `ref_id` to focus on specific sections (e.g., the</p><p>  response region or the citations dialog).</p><p>- Inline citations like "Wired +1" or "Meta +5" indicate the brand is cited multiple times</p><p>  within the response — count these as separate citation signals when tallying frequency.</p><p></p> |

## Social Links Footer
Thank you for spending time here. ❤️<br>Got questions, ideas, or feedback? Find me on <a href="https://www.linkedin.com/in/yzsun/" target="_blank" rel="noopener noreferrer">LinkedIn</a> and share what's on your mind.