Browser Use 60K stars. Manus $5B. Comet & Sky from incumbents. The horizontal race is closed but vertical wedge is wide open.
AI browser and web agent is the single hottest category in the 2026 AI map. The market sized at $4.5B in 2024 is projected to $76.8B by 2030 at 32.8% CAGR. The headlines are loud: Browser Use crossed 60K+ GitHub stars and became the default open-source agent harness; Manus (the Chinese team led by Yichao "Peak" Ji) hit a $5B valuation pre-IPO; Zhipu shipped AutoGLM as a native agent on China devices; Perplexity launched Comet as an agentic browser; The Browser Company shipped Dia/Sky as the consumer agent answer. Every foundation model lab (OpenAI computer-use, Anthropic computer-use, Google Project Mariner) is now also a browser-agent vendor. The horizontal "general purpose agent" race is closed — you cannot ship a 19th general agent and win. But the vertical workflow wedge is wide open: an agent that does one job (book a hotel, fill a tax form, file a benefits claim, run a Shopify SEO audit, automate Reddit research, scrape and reconcile competitor pricing) at $50-500/month is a clean indie business with near-zero competition from horizontal platforms. The 2026 indie playbook here is to pick one workflow, ship it on Browser Use or Playwright, and own a niche before Manus or Comet generalizes into it.
Founded by Magnus Müller and Gregor Zunic in Zurich. Wedge: clean Python harness that lets any LLM drive a real browser. The default OSS substrate every indie agent now runs on. Distribution = GitHub virality.
Chinese team led by Yichao "Peak" Ji. March 2025 launch went viral as "the autonomous task agent." Sustained ARR but stuck on activating Western enterprise. The lesson: viral demo to monetized B2B is non-trivial.
Zhipu AI's mobile-first browser agent. Deep OS integration on China Android. The China answer to OpenAI computer-use, distributed via OEM partnerships with Honor, OPPO, Xiaomi.
Perplexity's bet that the browser becomes the agent UI. Free + leverages the search distribution. The risk for indies: when Comet adds workflow X, your indie product loses a billboard of distribution.
From the Arc browser team. Pivoted from Arc to agentic browser in 2024. Consumer-first, design-led, lacks B2B revenue. Strong taste, unclear monetization. The team to watch.
Founded by ex-Baidu execs. Wedge: agentic search that does multi-step research and produces sparkpages. Strong in Asia, growing in US prosumer. ~$30M ARR estimate end 2025.
Anthropic shipped computer-use in late 2024. The API every serious indie agent now uses. Computer Use 2.0 (2025) made long-horizon tasks reliable enough for production. Compete with indies as infrastructure, not as a product.
Founded by Jesse Zhang. Picked one workflow — customer support automation — and dominated. The case study for vertical agent strategy: pick one job, sell it for $50-200K ACV, eat the BPO market.
The indie win in this category is operator-led. If you've manually reconciled competitor pricing across 12 marketplaces, filed 200+ benefits claims, or run weekly Reddit research for 3 years, you know the failure modes the foundation model labs never see. That domain knowledge is the moat.
Browser Use + Anthropic computer-use means a serious vertical agent now ships in 2-4 weeks of focused work. If you have engineering velocity and one clear workflow, you can beat the well-funded Series A team to market because they spent 6 months on "general purpose."
The indie sustainable model in this category is mid-priced B2B. Solo Realtors, Shopify operators, recruiters, paralegal teams all pay $200-1,500/month for a workflow that saves 5 hours/week. Consumer ($20/month) is dominated by Comet and Manus and is not where indie wins.
The horizontal general agent race is over. Manus has the brand, OpenAI/Anthropic have the model, Perplexity has distribution. A generalist agent from a 2-person indie team has zero shot in 2026. Specialize or die.
Web agents that scrape Amazon, LinkedIn, Twitter, Booking via DOM walk break every 60 days when the site ships A/B. If your product has zero abstraction over site brittleness — failure logs, retry trees, alternative paths — you're shipping a maintenance nightmare.
$20/month consumer agents are the Comet/Manus arena. Their CAC math beats indies by 10x because they own search distribution. If your business plan needs millions of $20 consumer subs, you cannot win this category.
Operator with deep workflow + 1 strong engineer
Industry vet + technical co-founder
Eng founder with API + integrations DNA
This is your best-fit track in 2026. Browser Use + Anthropic computer-use = ship in 30 days. If you have one workflow you obsess over, you can hit $20K MRR before any VC notices. The horizontal players cannot follow you into a niche.
The industry-bundled agent platform is closed without you. Engineering-only teams cannot name the 12 daily workflows of a recruiter or paralegal. Your industry knowledge plus one strong engineer is a $5M ARR business in 24 months.
If you build prosumer agents and have audience, you can launch on X / Reddit and get to $5K MRR in 60 days. Less defensible than vertical B2B but faster to first revenue.
5 min · 12 questions · Free · Get your archetype + top 3 matching tracks
Take the quiz →