Pomegra
Reduck, a French-American startup, closed a $3M pre-seed round led by Daphni to give AI agents ready-made browser scripts instead of step-by-step clicking.
Key Takeaways
- Reduck raised $3M in a pre-seed round led by Daphni, announced on 22 October 2026.
- Drysdale Ventures and Galion.exe also joined; valuation was not disclosed.
- Its library holds 1,200+ scripts across 250+ websites; the team will grow to eight.
Lead
Reduck, a French-American startup building web automation for AI agents, announced a $3M pre-seed round on 22 October 2026. Daphni led the financing, with Drysdale Ventures and Galion.exe participating. The company did not disclose a valuation.
The money will fund hiring in AI engineering, full-stack development and infrastructure. It will also go toward expanding the script library and launching a developer offering. After the round, the team will number eight.
What Does Reduck Actually Build?
Reduck builds Reduck MCP, a platform that lets AI agents find, run and create reusable browser automation scripts. An agent calls a prebuilt script as a tool, instead of working out each page and click from scratch.
The library currently holds more than 1,200 ready-to-run scripts covering over 250 websites. Reduck says the approach makes agents about three times faster and cheaper on average. Tasks it cites include downloading invoices and running LinkedIn market research. That figure is the company's own and has not been independently verified.
The product targets a specific gap. Many websites and business applications offer no API and no MCP interface, which forces agents to drive a browser the way a person would. That route is slow and costly, since each step consumes model calls.
Who Is Behind the Company?
Reduck was founded by Daniel Huynh (CEO), Louis Abraham (CTO) and Raphaël Millet (COO). Huynh and Millet co-founded Mithril Security, which was acquired by H Company. Huynh also helped launch LaVague, an open-source AI web agent that reached about 6,000 GitHub stars.
Abraham is an AI engineer who co-founded the encryption platform Secrecy and previously did biomedical machine learning research at Stanford. The team therefore arrives with direct experience of the approach it now says is too slow: LaVague agents navigated sites step by step.
Why Does a Pre-Seed Round Draw Attention Now?
The round matters because browser-driving agents have a cost problem that scripts could ease. Every exploratory click costs tokens and time, and the same task is often repeated by many users. Caching a working path as a script turns a repeated expense into a one-time one.
Early customers described the effect in concrete terms. A venture investor at Mox Venture cut monthly invoice retrieval from five hours to thirty minutes. A BentoML user reduced a Sales Navigator profile export from an afternoon to two minutes. A Hyperlex user cut prospect account setup from hours to minutes. These are testimonials published by the company, not audited results.
The skeptical reading is that $3M is a modest bet and the moat is unproven. Scripts break whenever a site changes its layout. A library of 1,200 scripts is only an asset if it is maintained cheaply, and Reduck's plan to have AI agents write and repair scripts autonomously is aimed at exactly that problem. Browser-agent vendors and the sites themselves could also narrow the gap by shipping official APIs or MCP servers.
What Comes Next for Reduck?
Reduck plans a developer offering for building custom agent systems and automating knowledge work. It is also developing agents that create and maintain automation scripts without manual effort. The company positions the offering around what it calls generative engine optimization, though it has not detailed how that works.
The next test is coverage and reliability. Growing from 250 sites to several thousand, while keeping scripts working through site updates, would show whether the model scales. A pre-seed round of this size gives the team room to find out, but not a long runway for broad expansion.
Outlook
Reduck enters the crowded AI agent tooling market with a $3M pre-seed, a Daphni-led cap table and a founding team with prior browser-agent experience. Its pitch rests on one claim, that reusable scripts beat step-by-step navigation by roughly three times on speed and cost. Whether that holds as the library grows, and as sites change under it, will decide if the company moves to a larger round.
