How AI tools can save you 11 hours per week

Most “LLM experts” tell you to optimize for AI search, without breaking down how LLMs actually work. The truth: LLMs don’t crawl the web like a search engine does. They only do live searches when queries demand freshness. Perplexity is the outlier. It crawls Bing index in real-time, and provides cited answers. Here’s a no BS-guide on how LLMs actually work: 1. Many prompts don’t trigger a live search (only queries that trigger freshness) 2. LLMs don’t do massive web searches (footprint) for live searches 3. Most LLMs use Bing (except Grok and Gemini) as their default source of truth 4. LLMs don’t crawl Reddit. They consider Reddit as an authoritative source if it appears in SERP results for prompts. We did a deep dive last month where we shared the exact process we use to increase AI visibility for clients + real tactics that work today. Want the recording? Comment "LLM" and I'll send it to you right away.

For some time, there has been consensus that both backlinks and mentions are important search signals for LLMs, arguably as important as they are to Google. New analysis of 35,000 data points by Kevin Indig (his Substack is excellent) has confirmed this, and surfaced some clearer findings worth noting. When it comes to backlinks and LLMs: 1. Authority helps, though only once you cross thresholds: visibility gains aren’t linear. 2. Quality outweighs volume: diverse domains matter more than raw totals. 3. Nofollow links carry weight, often equal to dofollow links in AI visibility. 4. Image links can influence mentions, sometimes more strongly than text links. The implication: backlink strategies optimised purely for Google don’t always translate into AI-driven visibility. Authority, diversity and link type all play different roles in how models surface brands. Backlink strategies should consider the referring-domain-to-backlink ratio, factor in image backlinks, and treat nofollows as meaningful signals. In AI search, it’s not just about getting more links: it’s about getting the right mix.

Tracking AI-driven search queries isn’t straightforward yet. Google hasn’t officially shared how to see queries showing up in AI Mode, Gemini, or Google AI Overviews. But here’s a simple trick using Google Search Console. We can filter question-style queries first using regex: ^(who|what|where|when|why|how|was|did|do|is|are|aren't|won't|does|if)\b Then, we can focus on longer queries with a word-count regex: ([^\"\s]*\s){8,} 8 is the minimum number of words; you can increase it to 10 or 12. And finally, combine both filters to get long question-style queries: ^(who|what|where|when|why|how|was|did|do|is|are|aren't|won't|does|if)[\"\s].*([^\"\s]*\s){10,} How to apply in GSC: Go to Performance > Search results > + New > Query > Custom (regex) Paste your regex, run the report, and you’ll see the longer, question-style queries your audience is actually asking. This simple trick can help SEOs start analyzing queries that might appear in AI-driven search results.

My first actual LinkedIn post in 10 years, I think? Breaking tradition for a small announcement... I've made a free tool for researchers to use to tag survey responses using AI. If you have large datasets from NPS or CSAT, or if you run your own surveys with free text questions, then there is a gold mine of information in that data - waiting to be classified and tracked. And now it doesn't take literally forever to do by hand! What is it? - An open source tool for enriching survey data through custom tags - Specify the tags you're interested in, giving the AI as much context as you need to help it tag accurately based on your interests - Choose from a huge number of models (GPT5, Gemini Pro 2.5, etc.) - Get a neatly tagged CSV out at the end that annotates every row with the themes and tags you've specified - Bring your own API key, no subscription and your key never stored Please shout if you have any questions or feature requests. :) Link here -> https://data-tagger.com

🚨 Google just changed how “Show More” works inside AI Overviews… and it’s a big deal. Instead of expanding more text and citations, the new version jumps users straight into AI Mode. That means: - Fewer clicks to publishers - Less visibility for original sources - And another quiet shift toward AI-first SERPs This isn’t just a UI tweak, it’s a signal. Google is training users to stay inside its AI ecosystem instead of visiting your site. 👇 In this carousel, we break down: - What this update actually does - Why it’s dangerous for site owners - And tactical moves (from the PressWhizz playbook) to defend your traffic & authority If you rely on search traffic, this matters. If you build with data, this is your early warning. ⚙️ Adapt now or watch AI eat your CTR.

Your Google Search Console already knows what people are prompting in AI mode. You just haven’t looked yet. Here’s a small tutorial to find out. Simple. Free. Straight from your Google Search Console. Steps: 1️⃣ Open Google Search Console 2️⃣ Go to the Performance tab 3️⃣ Click Filter → Query 4️⃣ Select Custom (Regex) Match 5️⃣ Use this Regex that finds queries with more than 10 words ^(\S+\s+){10}\S+ 👉 Tadaaa! You’re now looking at the long prompts people are typing — likely the ones used in Google’s AI Overview or AI mode. Use this data to: ✅ Understand which AI searches your site appears in ✅ Identify “prompt-style” queries that mention your topics ✅ Refine your content strategy for both new and existing pages 💡 Real SEO advantage starts with understanding how people talk to AI, not just search engines.

Imagine working with cheat codes turned on. That’s what the right AI stack feels like. Not more hours. Not more stress. Just… shortcuts that make you look 10x faster and sharper. The truth? Most people treat AI like random apps. They try one here, one there, and then complain it doesn’t “fit” into their work. But when you stack the right tools together, it’s like switching from normal mode to God mode. Here’s my cheat sheet 👇 - OpenAI → The genius mode cheat Drafts your ideas, codes your logic, and brainstorms faster than your brain can type. - Otter.ai → The mind reader cheat Every meeting, every conversation recorded, transcribed, searchable. Never “miss that point” again. - Runway → The movie magic cheat Turn a simple prompt into studio-level visuals and videos. Think Hollywood, minus the budget. - Airtable → The memory palace cheat Messy data? Scattered docs? Airtable turns chaos into systems that remember everything. - Google Trends → The crystal ball cheat It shows you what the world will care about before the world itself knows. Perfect for spotting the wave before it peaks. This stack isn’t about tools. It’s about leverage. It’s how one person can do the work of ten without burning out. 👉 Now I’m curious, if you had to pick ONE “cheat code” tool in your workflow… what would it be?

OpenAI just released Atlas. 6-month window to own your market. Then it’s over. What changed: Atlas doesn’t send customers to websites anymore. It gives them the answer. Recommends the business. Done. Customer converts inside the browser. Your website? Never seen. Right now: Competitors have zero AI citation strategy. Still burning cash on Google rankings. Completely asleep. This is your window. The timeline is brutal: Months 1-3: AI builds trusted source database Months 4-6: Patterns solidify Month 7+: Game over for late movers Ahrefs data: #1 Google rankings lost 34.5% of traffic to AI. Lower-ranked pages cited by AI won instead. Rankings don’t matter anymore. What happens: Competitor ranks below you. Atlas cites them as expert. Customer never sees you. You lose without knowing. Two types of businesses forming RIGHT NOW: AI recommends them. Or customers never see them. Pick one. What winners do immediately: Content for AI citations, not keywords. Expert credentials AI verifies. Structured data AI reads. Fresh intelligence AI trusts. Every day you wait: Competitor builds AI trust. Moat gets deeper. You fall further behind. Early movers own this forever. Late movers wonder where their customers went. This isn’t advice. It’s a warning. Move now or hand your market to faster competitors. Taking action this week or watching others dominate your market?

You can't build your house on Google search alone anymore. If a quarter of your potential customers are getting fast, ad-free, AI-generated answers... are you even being seen? And what will that number be in 3 years? Here's your action plan to future-proof your business: Check Your Traffic: Go to Google Analytics. Is AI search already sending traffic to your site? You need to know your current baseline. Optimise for AI: Just like you optimise for Google, you need to start thinking about how to be the cited source for AI. (Hint: Think clear, structured, authoritative content that directly answers questions.) Take Action NOW: Awareness is step one, action is step two. Don't let your competitors become the default AI answer!

Forget surface-level summaries. You want synthesis with logic, nuance, evidence. Google Gemini can help you with it. Here's how you can create deep research reports with Google Gemini 2.5 Flash. 1. Go to gemini.google.com 2. Select the 2.5 Flash model 3. Select ‘Deep Research’ in the ‘Tools’ section in the prompt box. 4. Write your prompt 5. You’ll get the deep research report prepared 6. Follow up if you need more information How to use the Deep Research report? - Share with a custom link - Share on social media platforms - Export to docs - Copy and use it somewhere else Pros: - Scans hundreds of sites and synthesizes info. - Produces multi-page reports with reasoning steps. - Lets you upload files to guide research. - Flash model balances speed and efficiency. - Saves hours on deep, complex research. Cons: - Can over-analyse and drift off-topic. - Still prone to hallucinations. - Reasoning may be manipulated. - Slower on heavy queries. - Privacy risks with uploaded files. Do's: ✅ Keep prompts clear and simple. ✅ Refine the research plan upfront. ✅ Ask follow-ups for deeper insights. ✅ Upload files for context. ✅ Export results to Docs or audio. Don'ts: ❎ Don’t expect instant results. ❎ Don’t overload unrelated tasks in one prompt. ❎ Don’t exceed usage limits. ❎ Don’t trust without verifying sources. To get a more detailed explanation,  Check the carousel below ��� What do you think of the deep research tool in Gemini? Comment below 👇 Learn AI for FREE: https://lnkd.in/eAAacxSK  ♻️ Share this to help your network create deep research reports with AI