The Reduck team, from the company's Oct. 22 funding announcement. Image: Reduck
Reduck has raised a $3 million pre-seed round to help AI agents use websites that have no API, the company announced on Oct. 22. Daphni led the round, with Drysdale Ventures and Galion.exe taking part, according to Reduck's blog and a report from Tech.eu. Daphni also posted about leading the round on LinkedIn.
The problem Reduck is going after is a common one. Plenty of sites and internal business tools offer no API or MCP server, so an agent that needs them has to click and type in a browser like a person would. Reduck's funding post calls that approach slow, expensive and unreliable. Every screenshot goes into the model's context, the company says, and the agent can take a different path each time it runs.
Scripts instead of clicks
Its product, Reduck MCP, gives AI agents a memory for the web: they discover, run and create browser automation scripts in the user's own logged-in Chrome, instead of clicking through websites step by step. Reduck maintains a library of official scripts for the most-used sites, and when none fits, the agent writes one and reuses it next time.
The company says that library currently holds more than 1,200 ready-to-run scripts across more than 250 websites. Its site lists ready-made scripts for LinkedIn, Twitter and Reddit, among others.
The speed numbers
In its funding post, Reduck says agents using its tool are on average 3x faster and cheaper across tasks like downloading 20 invoices from 5 vendors or doing market research on LinkedIn. A Sept. 29 post by CEO Daniel Huynh reported a separate, narrower benchmark: Claude with Reduck MCP finished three test tasks 2.5x faster than Claude with the Claude for Chrome MCP, and made 4.2x fewer tool calls. That post, which used Reduck's open-source Voyager eval framework, notes that "our sample is small." AgenticFeed has not tested these figures.
Who is behind it
Reduck was founded by Huynh as CEO, Louis Abraham as CTO and Raphaël Millet as COO, according to the company. Huynh and Millet previously co-founded Mithril Security, which was acquired by H Company. Reduck says Mithril launched LaVague, an open-source AI web agent.
The company says the new money has taken the team to eight people. Reduck says it is developing an agent swarm that creates and fixes scripts on its own. Next, the company says it is working on a developer offer for teams that need agents to reach systems without APIs, such as insurance and healthcare portals or internal CRMs, and its own apps for knowledge work, starting with generative engine optimization.
Source: Reduck blog, Reduck team, “Reduck raises $3M to build the web fast lanes for AI agents”
Additional sources: Tech.eu, Tamara Djurickovic, “Reduck secures $3M to make web automation faster for AI agents”; Reduck blog, Daniel Huynh, “Reduck MCP: the memory layer to make Agents 2.5x faster for browser tasks”; Reduck, “Company website”; Daphni on LinkedIn
