Reduck Raises $3M Pre-Seed to Make AI Agents Faster at Web Automation

Reduck, a French-American AI startup, has raised $3 million in pre-seed funding to make AI agents faster and cheaper at working across the web. Paris-based venture firm Daphni led the round, with participation from Drysdale Ventures and Galion.exe, according to Tech.eu.

The company is going after one of the least glamorous problems in agentic AI. Much of the software businesses rely on every day has no API or MCP interface, which forces agents to click through websites the slow way, screen by screen.

What Reduck Does

Reduck's core product, Reduck MCP, gives AI agents access to a library of ready-to-run browser automation scripts. Instead of reasoning through every page from scratch, an agent can find an existing script for the task, run it, or create a new one when none exists. The library holds more than 1,200 scripts covering over 250 websites and business applications.

The company says the approach makes agents roughly three times faster and cheaper on routine work such as downloading invoices or running market research on LinkedIn.

Reduck was founded by CEO Daniel Huynh, CTO Louis Abraham and COO Raphaël Millet.

That library is the product's real moat. Rather than treating every website as a new puzzle, Reduck treats common tasks as solved problems that can be shared, versioned and reused across customers, much like a package registry does for developers. Each new script makes the next agent run cheaper for everyone on the platform.

How It Will Use the Funding

The new capital will go toward hiring across AI engineering, full-stack development and infrastructure. Reduck plans to improve its automation technology, expand its script library and launch a developer offering for teams building custom agent systems, with insurance and healthcare portals and internal CRM tools among its early targets.

The startup is also building AI agents that create and maintain scripts on their own, which would let the library grow without a matching increase in headcount. It is exploring generative engine optimization (GEO) as well.

Why It Matters Now

The agent race has largely focused on model intelligence. Reduck is betting on the plumbing. Every time an agent has to visually navigate a portal, it burns time and tokens, and every small change to a website's layout can break the workflow. A shared, maintained library of scripts turns that one-off effort into reusable infrastructure.

The timing tracks with how companies are deploying agents in 2026. As businesses move from pilots to production, cost per task and reliability have become the metrics that decide whether an agent stays in the workflow. A tool that cuts both by a factor of three on common jobs is an easy pitch to engineering teams watching their inference bills.

The bigger question is defensibility. Automation scripts are only valuable if they keep working as websites change, so Reduck's long-term edge will depend on whether its self-maintaining agents can keep the library current faster than competitors can copy it. With fresh capital and a developer product on the way, the company is about to find out.

Source

Tech.eu — "Reduck secures $3M to make web automation faster for AI agents"