Discover the Best AI Tools

We test the top AI tools for writing, video, images & music — so you don't have to.

Explore AI Tools

AI Tools Every Roofer Should Use in 2026

Image
 Roofing used to be ladders, tape measures, and missed calls while you were up on a roof. In 2026, a lot of that work happens before anyone climbs anything — satellites and drones measure the roof, AI answers the phone while you're on a job, and software writes the estimate. Here are the tools working roofers actually use, job by job. Measure the Roof Without Climbing It Roofr turns satellite images into a full measurement report, and its Starter plan is free  — you only pay $13 to $19 per report. EagleView does the same job with aerial imagery and property data, with reports starting around $18. Hover takes a different route: snap photos of the house with your phone and it builds a measured 3D model, with projects from $29 to $69 and the first three free. Any of the three beats a tape measure on a 40-degree slope, and the report doubles as the number your estimate is built on. Run the Whole Job From One Screen Once the estimate is done, AccuLynx and JobNimbus take over — lead...

OpenAI Says ChatGPT Now Handles 1 in 5 Searches — Here's the Catch

 

ChatGPT vs Google Search cover

OpenAI logo


Google logo


OpenAI says ChatGPT has reached a striking milestone: its volume of 'search-like queries' is now roughly one-fifth of Google's. If that is right, asking a chatbot first is no longer a niche habit. It is a major way people look for answers.

The claim came from OpenAI consumer product lead Adam Fry, speaking to Big Technology's Alex Kantrowitz at an OpenAI-hosted dinner in New York on November 2, 2026; the report was published November 3. Fry also cited ChatGPT's scale: 1.2 billion weekly users, up from 1 billion in about 60 days, plus 250 million weekly voice users and 1.5 billion images created per week.

What OpenAI actually claimed

Fry was not saying ChatGPT has replaced Google. He was saying that more questions that once went to a 

search box now go to a chatbot instead. People ask for explanations, comparisons, plans and recommendations, and ChatGPT often searches the web behind the scenes before answering. That behaviour — a question that triggers a web search — is what Fry counted. The shift is happening as OpenAI reshapes its products, including ChatGPT and Codex, turning ChatGPT into a broader workspace, not just a question box.

The catch: how the number is counted

Here is the catch. This is OpenAI's own claim, with no published method, no absolute search count and no disclosed source for the Google figure. Readers are asked to trust a ratio without the numbers behind it.

The counting method also inflates the total. Fry counts every question that makes ChatGPT search the web as one search, so one conversation can count several times. A user who refines a question, asks a follow-up or changes direction can generate several 'searches' from a single need. That is not the same as one person running one Google search.

Why independent estimates are lower

Independent estimates run lower, which is a warning against treating 20% as settled. SEOClarity put the share at about 7.5% in October 2026, while Ahrefs estimated about 12% in February 2026. Different methods, dates and definitions explain part of the gap, but the direction is consistent: outside OpenAI, the figure looks smaller.

There is another distinction. Query volume is not the same as visits sent to websites. A chatbot can answer a question without sending the user anywhere, which matters greatly to publishers. Google, meanwhile, is adapting its own search — AI Mode earns less than classic search in the company's own telling — while Gemini gives Google a chatbot of its own.

The take

None of this makes Google irrelevant, and none of it makes OpenAI's claim meaningless. The sensible reading is narrower: chatbots have become a serious first stop, especially when people want a direct explanation rather than a list of links. Until OpenAI publishes its method and counts, 'about one in five' should be read as a claim about momentum, not a precise measurement. The habit is real. The exact share is still disputed.

Comments

Popular posts from this blog

AI Tools Every Accountant Should Use in 2026

Top 5 Best AI Video Generation Tools in 2026

7 Best AI Website Builders in 2026