Rindler

Rindler is a Y Combinator-backed web automation platform founded in 2025 by MIT classmates Michael Serrano and Arthur De Los Santos, who conducted research at MIT CSAIL. It targets the portal-heavy busywork that teams still do by hand — logging into supplier portals, ATS boards, payer portals, bank sites, and government registries — and runs those tasks automatically, including behind login walls, on sites that have no API and never will.

What It Is

Rindler is a browser automation service that signs into the websites your team already uses and performs repetitive tasks on your behalf. Unlike RPA tools that record exact click paths and break when a page redesigns, Rindler works from what a screen means rather than where a button sat. You describe the job in plain language — "pull the open orders from our supplier portal every weekday morning" — and Rindler handles the login, navigation, data extraction, and scheduling. Results come back as structured rows, not screenshots, ready to feed into a spreadsheet or downstream system.

How the Workflow Works

The setup follows four steps:

• Describe the job — plain words, no builder, no flowchart
• Connect the site — sign in once; the session is encrypted and scoped to that one site
• Monitor the run — every step and every screen the agent saw is logged
• Scale with automations — save tasks and put them on a schedule; they run on their own

The first pass through a new site is where the mapping work happens. After that, the same task repeats without rediscovering the site. When a portal changes its layout, Rindler repairs the route centrally for everyone on that site — no ticket required, no charge for the fix.

Target Industries and Use Cases

Rindler is built for teams whose day is spent inside login-walled systems they do not control. The platform covers six portal-heavy verticals:

• Procurement & AP — supplier portals, invoice upload, PO status, vendor updates
• Government & Courts — registries, permits, court e-filing, records retrieval
• Healthcare & Insurance — prior auth, appeals, attachments, credentialing upkeep
• HR & Recruiting — work across every client's ATS, job posting, candidate pulls
• Banking & Lending — statement downloads, transaction exports, payoff quotes
• Commerce, Travel & Property — storefronts, booking, listings, calendars, reorders

The platform also supports one-off chat-style tasks on public sites like Greenhouse, Lever, Ashby, and state business registries.

Security and Session Model

Rindler's security model is designed around not holding passwords. For most sites, users sign in themselves on the real login page inside a browser Rindler manages; Rindler then holds only the resulting session, encrypted with AES-256-GCM under a key scoped to the user's account. A limited opt-in option allows storing sign-in credentials per site. Sessions are revocable at any time, and revocation deletes the encrypted record. The platform uses Anthropic (Claude) and OpenAI for AI processing, with training opt-out and retention switched off. Rindler is not SOC 2, HIPAA, or PCI certified as of the time of writing, and states this explicitly.

Performance Claims and Architecture

Rindler's homepage benchmarks the platform against browser-use, an open-source browser agent, across 38 runs on two grocery tasks in early 2026. According to Rindler's own published figures, using a pre-mapped site resulted in 4× faster task completions, 3× lower cost per run (due to returning structured records rather than full page DOM), and 2× fewer failed tasks compared to an unmapped agent. The MCP integration exposes four core tools — start_session, dispatch_action, extract_content, and close_session — allowing AI agents to call Rindler as a structured tool rather than scraping raw HTML.