A Browser Dies Nine Months After It Was Meant to Change Everything
When OpenAI launched ChatGPT Atlas in October 2025, it was pitched as a challenger to Chrome and a new front door to the web. Nine months later it is gone. According to TechCrunch’s report on the shutdown, OpenAI announced on July 9, 2026 that it was discontinuing Atlas — and 9to5Mac confirmed the browser was “scheduled to stop working on August 9”, with bookmarks that “will not transfer automatically” and had to be exported as an HTML file before the deadline.
The reasoning is the interesting part. TechCrunch frames the decision bluntly: “After a few months of experimenting, OpenAI appears to have concluded that the browser is a feature, not the destination.” Rather than kill the capability, OpenAI moved it — the agentic browsing functions migrate into the ChatGPT desktop app, which now includes a built-in browser and “computer use” capability, plus a recommended Chrome extension. Atlas the standalone product died; agentic browsing the feature did not.
Comet’s Opposite Bet: Free, Everywhere
While OpenAI concluded a standalone AI browser was not worth maintaining, Perplexity has been running the opposite experiment. Its Comet browser, which launched as a premium product gated behind a $200-per-month Perplexity Max subscription, was made free. Per TechRepublic’s coverage, Comet became available “for free, at perplexity.ai/comet” on October 6, 2025, in a worldwide rollout, dropping the $200 monthly price entirely.
The pitch is agentic: TechRepublic describes Comet’s assistant as able to “summarize pages, organize tabs, help draft emails, compare shopping options, and even manage tasks in the background.” Perplexity’s strategy is distribution over margin — give away the agentic browser to win users and habits, and monetize elsewhere. That is a defensible bet only if the browser becomes the daily surface where people run AI tasks, which is precisely the premise OpenAI just decided was too weak to build a standalone product around.
Two of the best-funded players in AI reached opposite conclusions about the same category in the same quarter. That divergence is the story.
The Consumer Pull Problem
The tension between those two bets exposes a real weakness in the mass-market agentic-browser thesis, and the clearest statement of it comes from someone building in the space. Kevin Jiang, a former Perplexity engineer who worked on Comet, launched a new AI browser called Polar aimed at knowledge work, backed by a $5.7 million seed round led by Madrona Venture Group.
Jiang’s diagnosis of why consumer AI browsers struggle is direct: “If you think about end-user consumers, they don’t book a flight or make a reservation every day… There wasn’t a strong pull for mass consumers to go to an AI browser and find value in it.” His answer is to target knowledge workers in sales, recruitment, marketing, research and operations — people who run repetitive multi-step web tasks all day — rather than the general public. Polar ships as a free browser with paid tiers starting at $20/month for higher AI task credits, letting users schedule workflows, save prompts and assign tasks to agents based on open tabs.
That repositioning — from “a better browser for everyone” to “an automation surface for professionals” — is the direction the whole category is bending. As TechCrunch notes, 2025 saw failed standalone attempts, and 2026 is now focused on “agents and automation,” with startups like Strawberry, Browser Use and Aside building in the space.
Advertisement
Why the Category Is Repricing So Violently
Prices are swinging from $200/month to free to $20/month across the category in under a year because nobody yet knows what an agentic browser is worth. The technology can genuinely fill forms, complete multi-step tasks and operate the web semi-autonomously — but the willingness to pay depends entirely on whether it becomes a daily habit or a niche power tool.
OpenAI’s retirement of Atlas is a vote for “niche, so fold it into the main product.” Perplexity’s free worldwide Comet is a vote for “land grab first, monetize later.” Polar’s professional focus is a vote for “the value is real but only for people who automate work all day.” All three can be partly right, and the shakeout underway is the market discovering which framing survives. For users, the immediate consequence is that powerful web-automation capability is now available at or near zero cost — a windfall that will not necessarily last once the category settles.
What This Means for Algerian SMEs
Free, capable agentic browsers arrive at a moment when Algerian small businesses could use exactly the kind of leverage they promise — provided the security tradeoffs are handled deliberately.
1. Pilot agentic browsers on repetitive back-office tasks first
The clearest value is in high-frequency, low-judgment web work: data entry across portals, form-filling, price comparison, lead research. Start a pilot on one such workflow, measure hours saved, and expand only where the automation is reliable. This is where a small Algerian firm gets the biggest return for the least risk.
2. Treat browser agents as a new attack surface, not just a productivity tool
An agent that can fill forms and act on your behalf can also be steered by malicious page content or leak credentials if misconfigured. Restrict what accounts and sites a browser agent can touch, keep it away from banking and sensitive systems until vetted, and require human confirmation for any irreversible action. The security questions the category reopens are not hypothetical.
3. Do not build a business process on a free tier that may not survive
Comet is free and Atlas was retired in the same quarter — pricing and availability in this category are unstable. If an agentic browser becomes load-bearing for your operations, plan for the day its terms change: keep the underlying workflow documented and portable so you are not stranded when a vendor reprices or shuts down.
Where the Shakeout Leaves Buyers
The agentic-browser category in 2026 is a live demonstration of a technology outrunning its business model. The capability is real and, for now, cheap: OpenAI, Perplexity and a raft of startups have collectively pushed powerful web automation to free or near-free pricing while they fight over what it is ultimately worth. The losers so far have been the standalone-browser bets — Atlas most prominently — and the survivors are converging on either giveaway-scale distribution or professional automation niches.
For an Algerian SME or freelancer, the practical read is simple: harvest the current windfall of free, capable web-automation tools to compress repetitive digital work, but architect around the volatility. Use these agents where they save real hours, gate them hard on security, and never let a free tier become a single point of failure. The shakeout will eventually pick winners and reintroduce prices; the smart move is to bank the productivity now without betting the business on any one product surviving it.
Frequently Asked Questions
What happened to OpenAI’s Atlas browser?
OpenAI announced on July 9, 2026 that it was retiring ChatGPT Atlas, and the standalone browser stopped working on August 9, 2026 — less than a year after its October 2025 launch. OpenAI concluded “the browser is a feature, not the destination” and moved its agentic browsing capabilities into the ChatGPT desktop app and a recommended Chrome extension. Atlas bookmarks did not transfer automatically and had to be exported as HTML before the deadline.
Is Perplexity’s Comet browser really free?
Yes. Comet launched as a premium product gated behind a $200-per-month Perplexity Max subscription, then became free worldwide on October 6, 2025. Its assistant can summarize pages, organize tabs, draft emails, compare shopping options and manage tasks in the background. Perplexity’s bet is distribution first, monetization later.
Are agentic browsers safe to use for business tasks?
They introduce new risks. A browser agent that can fill forms and act on your behalf can also be manipulated by malicious page content or leak credentials if misconfigured. The prudent approach is to restrict which accounts and sites an agent can access, keep it away from banking and sensitive systems until vetted, and require human confirmation for irreversible actions. The productivity is real, but so is the expanded attack surface.
Sources & Further Reading
- OpenAI Is Shutting Down Atlas, but Its AI Browser Ambitions Are Still Growing — TechCrunch
- OpenAI Explains What Will Happen When ChatGPT Atlas Shuts Down This Weekend — 9to5Mac
- Perplexity’s Comet AI Browser Is Now Free for Everyone — TechRepublic
- Perplexity Employee Who Worked on Comet Launches an AI Browser Aimed at Knowledge Work — TechCrunch
