Comet Browser vs Zapier AI Actions 2.0

“The category here is AI browser agent, and the direct competitors are Arc with Browse, Chrome's built-in Gemini integration, and every Playwright-wrapper startup that launched in 2024. The specific scenario where Comet breaks: any website with a CAPTCHA, a bot-detection layer, or a dynamic login flow — which is most of the websites people actually need agents to navigate. My 12-month kill prediction: Google ships Gemini-native agentic browsing into Chrome for free and Comet's entire distribution thesis evaporates. To earn a ship, Comet needs to demonstrate a reliable task completion rate above 80% on a published, third-party benchmark — not a cherry-picked demo on a frictionless checkout flow.”

“Direct competitors are Make.com's AI scenarios, n8n's agent nodes, and Microsoft's Power Automate with Copilot — Zapier wins on breadth of connectors but loses on price-per-task at scale, which is exactly where agentic workflows go sideways because agents are chatty and task counts explode unpredictably. The specific scenario where this breaks: any workflow requiring reliable state persistence across long-running jobs, or anything that touches data that needs auditability — the retry-and-branch model is fine for 'send a Slack message if this fails' but not for 'reconcile 10,000 invoice records.' What kills this in 12 months isn't a competitor — it's Zapier's own per-task pricing colliding with agentic loops that can burn through a monthly plan in an afternoon. That said, for the SMB user who just wants their CRM to auto-update from email and Slack, this is genuinely the path of least resistance.”

“The thesis Comet is betting on: within 3 years, the browser's primary interface is intent-driven rather than URL-driven, and the agent layer sits below the UI rather than on top of it as an extension. That's a falsifiable, specific bet — and it's one I think is roughly on time, not early. The second-order effect that matters here isn't faster form-filling; it's that Perplexity captures the session-level data that Google currently owns through Chrome, which fundamentally shifts who can build the best personal web model. The dependency that has to hold: agent reliability needs to hit a threshold where users trust it with consequential tasks, not just toy demos, and that threshold is further out than Perplexity's beta launch implies.”

“The thesis Zapier is betting on: within two years, the dominant unit of software work for SMBs is not the app but the workflow, and whoever owns the integration layer owns the agent runtime — falsifiable because if model providers ship native cross-app orchestration (OpenAI already has Operators, Anthropic has computer use), the connector graph becomes less relevant. The second-order effect that nobody is writing about: if this works, Zapier becomes the credentialing and trust layer for AI agents acting on behalf of users, which is a radically more powerful position than 'automation tool' — enterprises will pay serious money for an agent that already has audited OAuth tokens for 300 enterprise apps. Zapier is late to the agent framework trend relative to pure-play entrants but uniquely early on the integration-as-agent-runtime trend, and that's the bet worth watching.”

“The buyer here is unclear in a way that matters: is this a consumer product funded by attention and ads, or a prosumer tool with a subscription model? 'Free beta with pricing TBD' is not a business model, it's a deferral, and for a company that's already raised at a multi-billion valuation, that deferral is a red flag. The moat problem is real — Perplexity's agent layer is only as defensible as its model quality and browser telemetry, and Google can replicate both with Chrome's existing install base. What would need to change: a clear pricing architecture that shows users pay for task completion or saved time, not for a browser they'll abandon the moment Chrome ships the same capability.”

“The buyer is the same person who already pays for Zapier — an ops manager or solopreneur who wants automation without hiring a developer — but the pricing architecture is the problem: agentic workflows are fundamentally unpredictable in task consumption, and Zapier charges per task, which means the unit economics for the user get terrifying fast when an agent retries, branches, and fans out across a dozen apps. The moat is real — 7,000 connectors with auth already handled is not something you replicate in a weekend — but the business model doesn't survive the agent paradigm intact; you can't charge per-task when the whole point of agents is that they take as many tasks as they need. Until Zapier ships a per-agent or per-outcome pricing model, this is a retention feature for existing users dressed up as a new product line, and that's not a business, it's a defensive move.”

“The job-to-be-done is clean and singular: complete a web task I would otherwise have to do manually. That's a real job, and most tools in this space make users context-switch between a chat interface and a browser, which is exactly the friction Comet eliminates by collapsing them into one surface. The onboarding question I'd need answered before moving this to a strong ship: does a user reach a completed task in their first 2 minutes, or do they spend that time granting permissions and configuring agent scope? The opinion the product needs to have — and may not yet have — is which tasks it's opinionated about doing well versus which it declines, because an agent that attempts everything and fails unpredictably is worse than one that does three things reliably.”

“The primitive here is a hosted LLM orchestration layer that uses Zapier's existing connector graph as its tool registry — that's actually a defensible technical choice, not just a rebrand. The DX bet is that you never write a tool definition or manage auth, because 7,000 connectors already exist and credentialing is already handled; for anyone who's hand-rolled LangChain agents and spent three hours debugging OAuth, that's real value. The moment of truth is whether branching logic and retry semantics hold up on real workflows with partial failures — the docs show the promise but I'd want to see error handling that isn't just 'retry three times and give up.' Calling this a ship because the integration graph is a genuine moat and they didn't just wrap GPT-4 in a Zap.”