Riveter

Riveter is an AI web data extraction and enrichment platform delivered primarily as an API. Users describe the data they need in natural language, and Riveter's agents search the live web, navigate to relevant pages, and extract and structure the results, returning a finished dataset through a single request. The product covers four main functions: building datasets of companies, people, places or products from a prompt with no input data required; automating web research and enrichment of existing rows; returning consistently formatted structured output for ingestion into code, tools and reports; and scheduling projects to monitor sources for changes.

The platform's API exposes enrichment, dataset building, scraping, quick search and a search-agent endpoint, with asynchronous "runs" that can be polled or delivered via webhooks. Riveter publishes official TypeScript/JavaScript, Python, Ruby and Go SDKs, an MCP server for use from AI assistants such as Claude, Cursor, Codex and Windsurf, and a web app in addition to the API. Documented limits include a default rate limit of 30 requests per minute per endpoint group and row caps of up to 10,000 rows per enrichment request using a saved enrichment ID (1,000 rows with an inline configuration). The company states it handles bot detection, proxy management, headless browsing and site changes on the user's behalf, and can read PDFs, images and third-party APIs alongside web pages, combining them into one structured output.

Stated use cases span data and engineering teams replacing custom scraper infrastructure, pricing and competitive research, sales and go-to-market lead list building and enrichment, operations monitoring, and finance and risk work including KYB and compliance research. Industry applications listed include e-commerce, marketing, industry classification and real estate.

Founding story

Co-founders Abby Grills (CEO) and Cody Watters (CTO) each spent five or more years at Gusto before moving to Middesk and Retool respectively. They describe repeatedly encountering the same problem across engineering, data, operations and growth teams: the information needed was available on the web but there was no reliable way to get it into a structured, usable dataset at scale. Grills previously led 0-to-1 products at Gusto, including time tracking and state payroll tax registration, and worked on Address Risk at Middesk.

Business model

Riveter sells software access to its data extraction and enrichment platform, usable in-app, via API, or through its MCP server. Pricing is per action in credits rather than per seat, with a free plan, pay-as-you-go with automatic top-ups, a self-serve plan, and custom enterprise pricing.

Usage-based credit pricing charged per action, across free, pay-as-you-go, self-serve, and custom enterprise tiers, rather than per-seat licensing.

Traction

The Y Combinator company page and launch post state that teams at OpenAI, Snap and Fermat use Riveter to build datasets, and the site carries named testimonials from staff at Zeffy, Roundabout Technologies, Fermat and Multiply. Riveter has shipped multiple product iterations, from a spreadsheet AI data assistant and a corporate-strategy copilot through parallel AI search agents to a general web data API, and publishes SDKs in four languages plus an MCP server.

Latest developments

Riveter's most recent launch positions the product as prompt-to-dataset generation, where a single natural-language prompt produces a list and layered enrichments scraped live from the web, available in-app, via API or through the MCP server. The company has released a second-generation API (v2.0.0) with unified run lifecycle endpoints, webhooks, and official TypeScript, Python, Ruby and Go SDKs, while keeping first-generation endpoints working as legacy paths.