Reduck secures $3M to make web automation faster for AI agents

Reduck, a French-American AI startup, has closed a $3 million pre-seed round led by Daphni, with Drysdale Ventures and Galion.exe also taking part.

The company is building software for a very specific bottleneck: AI agents can be eager, but many websites and internal tools still don’t offer APIs or Model Context Protocol interfaces. That leaves agents to poke around the old-fashioned way, by clicking and typing in a browser, which gets slow, expensive and fragile fast.

Reduck’s answer is Reduck MCP, a platform that lets agents discover, run and create browser automation scripts. The pitch is simple. Don’t make the agent grind through each page step by step if a script can get the job done faster and with less compute.

The company says that approach can make agents three times faster and cheaper on tasks like pulling invoices from several vendors or doing market research on LinkedIn. Its library already contains more than 1,200 ready-to-run scripts across more than 250 websites, and Reduck now wants AI agents to help build and maintain those scripts on their own.

That bigger system is meant to improve reliability and scale. Alongside it, Reduck is preparing a developer product so teams can build custom agent systems that connect to applications without APIs, including specialist portals in insurance and healthcare, as well as internal CRM systems. The company is also looking at knowledge work automation, starting with generative engine optimisation.