In January 2025, OpenAI launched Operator, a browser-controlling agent using a new Computer-Using Agent model. In July 2025, Perplexity shipped Comet to its $200/month Max subscribers. In September 2025, Atlassian announced it was acquiring The Browser Company — the team behind Arc and Dia — for $610 million in cash. In October, OpenAI launched ChatGPT Atlas, its own standalone browser. The same month, Perplexity made Comet free worldwide. By March 2026, Perplexity itself was valued at $20 billion. Five of the most aggressive AI companies in the world built or bought a browser in fifteen months. That is not five product decisions. That is one signal.

What The Pattern Says

The browser is the location where humans interact with the entire web — every SaaS app, every workflow, every form, every internal tool. For two decades it was a passive viewer. The agent revolution has revealed it to be the most strategically valuable interface layer in software. Whoever owns the browser owns:

  • The action surface. Every click, scroll, and form submission is observable. An agent embedded in the browser does not need API integrations to act on a customer’s behalf — it acts as the customer, on whatever website the customer uses.
  • The context graph. What tabs are open, what page the user just left, what email they were drafting — this is the most accurate real-time model of intent that exists. ChatGPT Atlas ships explicitly to capture this signal that the standalone ChatGPT app cannot see.
  • The distribution layer. Perplexity’s strategy of giving Comet away free worldwide is the second-cheapest customer acquisition cost in software, after open source. Browsers are sticky in a way that apps are not.

The defensive logic is symmetric. Google’s Chrome is the existing incumbent. If users move their primary work to an agentic browser owned by Perplexity, OpenAI, or Atlassian, the search query — which monetises Chrome — moves with them. Google’s response was to ship Gemini directly inside Chrome in September 2025. The browser war that everyone thought was settled in 2010 is now a live front in the AI war.

Why The Convergence Now

Three preconditions had to land in 2024-2025 for browser agents to become viable. First, multimodal vision models that parse a webpage from raw pixels (OpenAI’s CUA, Anthropic’s Computer Use) replaced fragile DOM-parsing approaches that broke on every UI change. Second, action-taking models capable of multi-step tool use without losing the plot — exactly the capability that took the foundation labs from 2023 to 2025 to operationalise. Third, the cost-per-action collapsed enough that taking thirty actions to complete one user task became economically tolerable. None of these were true in 2023. All three were true by mid-2025. The product pivot was a response to the capability arrival, not the cause of it.

Why It Matters

For founders building agentic products, the browser is the new platform layer. If your product requires API credentials to twelve services, you will lose to a browser agent that needs nothing because it is the user. If your product runs as a Chrome extension, you are now building on an interface layer whose owner is also building agents that compete with you.

For SaaS incumbents, the read is harder. The browser agent does not care whether your product has a great UI — it scrapes, parses, and acts. The defensible moats are the data, workflows, and identity layers that exist behind the UI, not the UI itself. The companies that have spent the last five years over-investing in front-end polish are about to discover that the front end is being commoditised by a layer above them.

For investors, the signal is the convergence itself. Five independent companies — different funding bases, different go-to-market motions, different geographies — converged on the same product wedge inside one year. That is the same shape as the Stripe / Ramp / Coinbase agent architecture convergence. When this many serious players ship the same shape inside the same window, the shape is correct.

The Charaka View

Manthan Intelligence’s signal-detection work uses a “convergent shipping” filter — when three or more well-resourced, independent teams ship structurally similar products inside a six-month window, the underlying problem and the solution shape are both validated by the convergence itself. The browser-as-agent pattern passed that filter in October 2025 with the Atlas / Comet-free / Atlassian-completes-acquisition trio. The next eighteen months will consolidate it — Chrome will defend; Atlassian will productise Dia for work; Perplexity will trade reach for monetisation; OpenAI will use Atlas to capture the context graph that makes the next ChatGPT model materially better than the last. Founders building agentic products should pick the browser they are betting on now — the platform owner you ship for in 2027 is being chosen this year.

This analysis draws on TechCrunch’s Operator launch coverage, CNBC’s reporting on the Atlassian acquisition of The Browser Company, Perplexity’s Comet free-worldwide announcement, and Yahoo Finance’s reporting on Perplexity’s $20B valuation. Human editorial oversight applied.

This analysis is informational and does not constitute investment advice, a research report, or a recommendation to buy, sell, or hold any security.

Charaka Notes by Manthan Intelligence. Subscribe