The Death of the Browser Tab and the Rise of the Answer

I had, in 2023, more than a hundred browser tabs open at any given moment. I have, in 2026, zero. The browser tab — the central unit of thinking on the internet for thirty-two years — is dying. The answer is replacing it.

I had, at my worst in 2023, more than a hundred browser tabs open at any given moment. They were a graveyard. They were a planning system. They were a punishment. I would open Chrome on Monday morning and the tab strip would scroll three rows across the top of the screen. There were tabs from 2019. There were tabs I would never close because I might, someday, need them.

I have, in September 2026, zero browser tabs open. The Chrome window I am writing this in has six pinned tabs (Gmail, the New York Times, my bank, Notion, a weather page, and a Slack reading space) and nothing else. The change happened, by my standard, faster than any other shift in my adult computing life. It happened between November 2022 and the spring of 2025. It happened because six AI products ate most of what the browser tab was for.

Below is the list. I have changed my mind about several things I used to believe about how the internet is supposed to work. I have opinions on what to skip. There is a brief history of how the browser tab went from a 1994 Netscape invention to a 2026 artifact. And there is, at the end, an honest summary — because anyone who tells you the browser tab is fine has not, in fact, noticed that they have not opened a search result in nine months.

The list

1. ChatGPT Plus, $20/month (since December 2022). The single most-used application on my computer. I open it, on a heavy day, fifteen times. I use it for everything from drafting emails to summarizing articles to debugging my partner's Python scripts to planning what to cook for dinner. The $20 per month is, by every measure, the best value in my productivity toolkit. It is also, by my standard, the reason I have no tabs.

2. Perplexity Pro, $20/month (since March 2024). The answer engine. Perplexity is, in 2026, the product I open when I want a researched answer instead of a generated one. It cites sources. It reads the sources. It writes the answer. Trip planning, recipe research, news digests, gift research, restaurant recommendations — all of it used to be twenty open tabs. Now it is one Perplexity query and a paragraph.

3. Claude Pro, $20/month (since February 2025). The writing partner. I write for a living. Claude is, by every measure, the AI that writes most like a thoughtful adult. I use it to draft long documents, to edit my drafts, to summarize research.

4. Google Gemini Advanced, $20/month (since November 2024). The Google answer. Gemini is, in 2026, the AI tied to my Gmail, my Google Docs, my Google Calendar. It is the AI I use to find an email from 2021 or to summarize a 60-page PDF. I keep it because of the integration with the Google stack.

5. Cursor, $20/month (since June 2024). The coding answer. Cursor is, in 2026, the AI-native code editor built on top of VS Code. My partner is a software engineer. Cursor is, by his description, "the reason I do not open StackOverflow anymore." If you do not code, skip Cursor.

6. Microsoft Copilot Pro, $20/month (tested late 2024, canceled February 2025). The Microsoft answer. Copilot is, in 2026, the AI built into Word, Excel, Outlook, and PowerPoint. I do not use Office. I subscribed for one quarter to test it and canceled. The $20 per month was, in my workflow, redundant with everything else.

7. Notion AI, $10/month (since 2023, canceled August 2025). The note-taking answer. I subscribed because the marketing told me it would make my notes smarter. The reality, by my standard, was an extra $10 per month for features I never used. The replacement — Apple Notes, which added its own AI features in 2024 — does, by every measurement, what Notion AI claimed to do, for free.

Total monthly AI spend: $110. Total annual AI spend: $1,320. The browser tabs I had at peak (a hundred-plus) cost me, by my estimate, two hours a day of cognitive overhead. The six products that replaced them cost $1,320 per year. The math is, by any standard, the only math that matters.

What I have changed my mind about

A few beliefs I held in 2022, that the last three years have killed.

I used to think the browser tab was the unit of thinking on the internet. The tab was, in 2018, the natural container for "I might need this later." I would open twenty tabs for a single trip. I would open thirty tabs to compare products. The tabs were, in the end, a system for not making a decision. The AI products replaced the tabs because they make the decision for me. The tab is not the unit of thinking anymore. The answer is.

I used to think Google search was the front door of the internet. Google search was, in 2010, the only sensible starting point for any question. In 2026, Google is one of six starting points, and not the one I reach for first. Perplexity is the front door for researched answers. ChatGPT is the front door for generated answers. Google is, in my workflow, the front door for local businesses and old email addresses. That is, in 2026, a smaller door than it used to be.

I used to think AI products were toys. ChatGPT in December 2022 was, by my standard, a chatbot with a personality. By 2024, the same product was writing my first drafts, debugging my partner's code, summarizing PDFs. By 2026, it is infrastructure. The AI skepticism of 2023 — "it makes things up," "it cannot reason" — was, by my standard, the skepticism of people who had not yet integrated the tools into their daily work. The integration is, in 2026, the only thing that matters.

I used to think I needed to read every source. The internet research model of 2018 — open twenty tabs, read every source, synthesize yourself — was, by my standard, the model of a graduate student with too much time. The model of 2026 is: ask Perplexity, check the citations, click through to the two sources that look most important, move on. The number of articles I read end-to-end has fallen by, by my estimate, 80% in three years.

I used to think browsers were different. Chrome, Safari, Firefox, Edge — the browser wars of the 2010s were real. The browser wars of 2026 are mostly about AI. Arc launched "Arc Search" in 2023. The Browser Company released "Dia" in 2025 — an AI-native browser where the address bar is also the chat bar. Google rebuilt Chrome around Gemini. The browser is no longer a window onto the web. The browser is, in 2026, a window onto an AI that browses the web for you.

What to skip when building your answer stack

If I were advising a friend who wanted to set up their first AI subscriptions in 2026, I would tell them to skip these entirely at the start.

More than two general-purpose AI subscriptions. ChatGPT Plus, Claude Pro, Gemini Advanced, Copilot Pro, Perplexity Pro — they all do, by my standard, 80% of the same things. The marginal value of the third subscription is, in most workflows, marginal. Pick two: one for writing, one for research. Skip the rest.

AI products that lock you into one ecosystem. Notion AI ties you to Notion. Copilot ties you to Office. Gemini ties you to Google. The lock-in is, in 2026, the most expensive part of most AI products. The lock-in is also the thing you will regret in two years. Pick AI products that live in their own apps.

AI products without citations. ChatGPT makes things up. Perplexity does not, by default, make things up. The difference is, in my workflow, the difference between useful and useless. For any research question — trip planning, product comparison, news digest — use the AI that cites. For any generated question — drafting, brainstorming, coding — use the AI that does not need to cite.

AI hardware. The Humane Ai Pin launched in November 2023 at $499. It was discontinued in February 2024. The Rabbit R1 launched in April 2024 at $199. It is, in 2026, a $99 device that does what your phone already does. The AI hardware revolution of 2023-2024 was, by my standard, a revolution of products that did not need to exist. The phone is the AI interface. Skip the pendant.

Anything that promises to replace Google. Google search is, in 2026, still the front door for local businesses, old emails, technical documentation, and the long tail of the web that AI products have not yet indexed. The "Google killer" framing is, by my standard, mostly marketing. The AI products do not replace Google. They replace the workflow around Google. The workflow was: open Google, click a result, read the article, repeat. The workflow is, in 2026: ask the AI, check the citations, move on.

A brief history

The browser tab is, by every measure, a 1994 invention. Marc Andreessen's Netscape Navigator 0.9 — released in October 1994, the first commercial browser — opened one web page per window. Microsoft Internet Explorer introduced tabs in 1997. The tab was, by every measure, a productivity hack for limited screens.

The tab became standard with the rise of Google Search in 1998 and the explosion of the web in the early 2000s. Google Chrome launched in September 2008 with one tab per process — a technical choice that made tabs crash-resistant and turned the tab into the primary unit of multitasking on the web. By 2010, the tab was the central metaphor of personal computing.

The 2010s were the golden age of the tab. The average user, by Chrome's own statistics, had 8-15 tabs open at any time. The power users had 100+. The browser extensions industry (OneTab, The Great Suspender, Toby) was, by my standard, an industry built around the cognitive failure of the tab. The tab was not a thinking device. The tab was, in most cases, an anxiety problem waiting to disappear.

The answer to the tab started with ChatGPT's public launch on November 30, 2022. ChatGPT reached 100 million users in two months — the fastest adoption of any consumer product in history. Perplexity launched its Pro tier in 2023 with the explicit branding of an "answer engine." Google launched AI Overviews in May 2024. OpenAI launched Operator in January 2025. Google launched full AI Mode at Google I/O in May 2025. The Browser Company launched Dia in 2025. The pattern was consistent. The product was not a better browser. The product was an answer instead of a search.

The 2026 web is being rebuilt around the answer box. The 1994 web was built on links. The 2010 web was built on tabs. The 2026 web is being built on answers. The tab will survive — for entertainment, for social, for the long-form sites we choose to support. The tab is dying for the workflow that used to require it: the open-twenty-tabs-and-synthesize-myself workflow. That workflow is, in 2026, a single AI query.

The honest summary

The browser tab is, by every measure, a thinking device that has been replaced by an answering device. The 1994 web was built on tabs. The 2026 web is being built on answers. The tab will survive — for Gmail, for the New York Times, for the long-form sites we choose to support. Everything else is, by my standard, being eaten by the answer box.

What works for me: two general-purpose AI subscriptions (ChatGPT for drafting, Claude for thinking), one research AI subscription (Perplexity, for cited answers), one Google AI subscription (Gemini, for the email stack), and one coding AI subscription (Cursor, for my partner). Total: $100 per month. Total tabs open: zero.

What does not work for me, and may not work for you: the idea that you need every AI product. The five-AI-subscription stack is, by my standard, a $100 per month redundancy tax. Pick two. Skip the rest.

Three years in, my browser has fewer tabs than at any point in my adult life. I have, by my standard, more answers than at any point in my adult life. The tab was a thinking device. The answer is, in 2026, a better one.