Disclosure: This article is based on the vendor’s public information and launch coverage checked on October 27, 2026. WebRun is a brand-new product and this is an early research review, not a long-term hands-on test. Some links on this site are affiliate links; see our Affiliate Disclosure page for details.
What is WebRun?
WebRun, at webrun.ai, describes itself as an AI business automation platform for business owners, startup founders and small teams. Its agents take on the repetitive work a team does every week, inside the tools that team already uses, rather than asking the team to move to a new system. WebRun launched its AI agents for business operations on 20 October 2026, according to its own announcement, and that launch also included a developer API.
The central idea is simple: each agent takes on one recurring job. The vendor gives examples across customer service, sales, finance and operations, such as answering customer emails, routing support tickets, replying to Google reviews, responding to new leads, following up on outstanding quotes, chasing overdue invoices, reconciling vendor bills, preparing recurring reports, booking appointments, sending reminders, copying new bookings into a schedule, and turning work-order emails into jobs.
Key features
The most important feature is that agents work in a real browser, clicking and typing inside tools such as Gmail, HubSpot, Salesforce, QuickBooks and Xero, so the vendor says there are no APIs to connect and no code to write. The vendor also lists more than 14,000 ready-made workflow templates, guides for more than 500 apps, and persistent cloud environments with saved browser logins, file storage, memory and scheduled tasks. For longer jobs, an agent can work through a worklist one item at a time and keep track of what is complete, what comes next and anything it skipped, with the reason.
The other feature that stands out on paper is control. WebRun says you decide which websites an agent can visit and which actions it may take, and its Automation Policies check each proposed click, keystroke and page visit before it happens, then allow it, block it or pause for your approval. When an action is set to require approval, the run pauses and asks you in Telegram or WhatsApp, and every run is recorded so you can review what happened. Developers can also use the same infrastructure through a REST API and an MCP server, listed in the official MCP Registry as ai.webrun/webrun.
How WebRun works
WebRun’s own homepage gives the flow in three steps: describe the job, approve the connection, and put it on a schedule. There are three ways to set up an agent: describe the job in plain language, record your screen while you do it once, or start from a ready-made workflow template. The agent then does the work in a real browser, in the cloud, on the schedule you set, so your own computer does not need to stay on. Jobs that cannot be undone, such as emailing a customer, can wait for approval, and the vendor’s example flow includes reading work-order emails in Gmail, creating the job in Tradify and notifying the team in Telegram.
WebRun pricing
Pricing is public on webrun.ai/pricing, checked on October 27, 2026, and it is priced in USD. Starter is $20 per month for about 2,000 agent tasks, with a 7-day free trial, 3 agents running at once, 5 environments, 50 active workflows and email support. Pro is $100 per month for about 10,000 tasks, 10 agents at once, 20 environments, 100 active workflows and priority support. Business is $200 per month for about 20,000 tasks, 50 agents at once, unlimited environments and active workflows, dedicated support and an audit log. Extra usage is billed from pay-as-you-go credits, at about 1 cent per task, and the vendor says plans can be cancelled at any time. What is not shown as a fixed public price on that page is a separate annual rate or a high-volume developer rate, which the vendor frames as a call or special rates, so check the pricing page before budgeting for those cases.
Who WebRun is for
WebRun is positioned for business owners, startup founders and small teams with operations-heavy, recurring work, and that positioning fits the examples it gives: a cleaning company keeping jobs staffed and confirmed, a team chasing invoices, or a small business replying to reviews and following up leads. It will likely appeal most to anyone living in browser-based tools such as Gmail, HubSpot, QuickBooks, Xero, Slack and Telegram, who can name a job that repeats every day or every week and is willing to set approval rules before letting an agent run it. Researchers with recurring browser-based collection or follow-up work could test the same pattern, but the vendor’s positioning is business operations, not research, so that use should be judged on the specific workflow, not on a research claim.
Who should skip it
Skip WebRun for now if you need proven reliability today, because a product launched on 20 October 2026 cannot yet show a long public track record, independent long-term tests or a large body of user reports. Be cautious with regulated workflows as well: WebRun does describe compliance presets and policy controls, but this early review cannot verify how those controls behave under a specific regulator, audit or data-handling rule, so test them against your own requirements before any sensitive rollout. Finally, if your core work lives in native desktop apps rather than the browser-based tools WebRun describes, this browser-agent approach is not the right fit, even though the vendor also offers developer infrastructure around cloud environments.
Pros and cons
Pros: it works inside tools teams already use, with no code and no new system to learn for each job; setup can start from a plain-language description, a screen recording or a template; approval pauses and recorded runs give a clearer audit trail than a fully silent agent; and the task-based pricing is easy to compare, starting with a 7-day trial on Starter. Cons: WebRun is a brand-new product with a thin public track record, so reliability, support quality and long-term pricing are still unproven; Pro and Business jump to $100 and $200 per month for larger allowances, and extra usage is pay-as-you-go; approval pauses can slow work that needs to run without a person in the loop; and the core pitch is browser-based, so it does not replace work that lives outside web tools.
Verdict
Early verdict: 6.5/10 – an intriguing take on browser agents, but wait for a track record unless you enjoy testing new tools. The idea is genuinely useful: let an agent do the same small job inside the same tools, on a schedule, with approvals for the risky steps. The score stays in the middle because this review is based on launch-week public information, not on months of use, and the product still has to prove that its agents stay reliable when websites change, that its policies work in real teams, and that its support and pricing hold up at scale.
For related reading on AI automation, see our Review Methodology and the Gumloop tutorial.