Magentic raises $18M to put AI procurement agents inside manufacturers

Magentic (London) has raised $18 million (~€15 million) in a Series A led by Felicis, with existing investors Sequoia Capital and The Westly Group following on. The round was announced 17 September 2026, fourteen months after the company launched in July 2025 with a $5.5 million seed (led by Sequoia), bringing total funding to $23.5 million. Founders Robin Van Aeken (CEO, ex-McKinsey) and Odhran O'Donoghue (CTO, ex-OpenAI) are building AI agents — they call them Mages — that run industrial procurement end-to-end inside a manufacturer's existing software stack.

The claim is not autonomous assistance with procurement. It is autonomous execution: a Mage selects suppliers, negotiates contracts, places orders, and clears invoices, communicating through Microsoft Teams and email the same way a human buyer does.

What a Mage actually does

Procurement at a large manufacturer is data-intensive, repetitive, and consequential. A single Global 500 company may process millions of purchase orders annually across hundreds of suppliers and dozens of ERP systems. A Mage reads procurement data, diagnoses inefficiencies, surfaces the best supplier options at the current market price, and executes the transaction — escalating to a human only when a decision falls outside its operating parameters.

Revenue grew 40× since the seed round. Three of the ten largest beverage producers are customers. One Magentic customer processes over one million orders per year through the platform. Reported outcomes: 2-5% procurement cost savings and a 60% improvement in data quality on sourced orders. These numbers, if they hold at scale, are material: a 3% saving on a $1 billion annual procurement budget is $30 million.

Robin Van Aeken: "The companies that build the best intelligence into every decision they make will be the ones that compound their competitive advantage."

Speed as differentiation

Magentic's story is partly about pace. A company that raised its seed in July 2025 and is closing a Series A in September 2026 — led by a new investor (Felicis) at presumably a substantially higher valuation — has demonstrated something. Sequoia backed the seed; the fact that they did not lead the A is not a signal of diminished conviction so much as a signal that Felicis moved first in a competitive process.

The US market is the expansion target. Magentic already operates across Global 500 companies but the Americas, where industrial manufacturing is heavily concentrated and procurement automation is a well-established budget category, is the growth thesis. The company's competitors — SAP Ariba, Coupa, Ivalua, and a cohort of AI-native procurement startups — are mostly better funded and more established. Magentic's bet is that being AI-native from the ground up, rather than AI-retrofitted, produces a meaningfully different product.

Whether large manufacturers will genuinely delegate procurement authority — not just visibility — to a system that is fourteen months old is the question this round is buying time to answer.