Magentic Raises $18M to Build AI Workers for Industrial Procurement

Magentic is an AI company building digital workers for procurement, supply-chain, and operational teams at large manufacturers. Its central pitch is not another standalone procurement dashboard: the company says its multi-agent systems can operate through the enterprise tools employees already use, including Microsoft Teams, email, and internal systems. 49

The company announced an $18 million Series A on September 17, 2026. Felicis led the round, while existing backers Sequoia Capital and The Westly Group also participated. 19 Magentic launched in July 2025 with a $5.5 million seed round, meaning the Series A arrived roughly a year after launch. 414

What Magentic sells: AI “digital workers”

Magentic calls its agents Mages. They are described as multi-agent systems intended to work alongside procurement teams and inside manufacturers’ existing software environments, including ERP systems. 412

The goal is end-to-end workflow execution rather than isolated analysis. According to reporting on the product, the agents are designed to support activities such as:

• evaluating buy-versus-build decisions;
• identifying and selecting suppliers;
• negotiating and administering contracts;
• managing orders; and
• processing invoices. 9

The platform is aimed at both indirect procurement and direct spend—the raw materials and other inputs used to make products. That distinction matters because direct materials procurement is closely tied to manufacturing production and supply continuity. 9

Why manufacturers are the target

Industrial procurement teams often work across fragmented enterprise technology, purchasing records, contracts, invoices, technical documents, and supplier communications. Magentic’s thesis is that AI agents can coordinate work across those systems instead of requiring teams to manually assemble the context for every decision.

CEO and co-founder Robin Van Aeken has framed the opportunity against AI-driven infrastructure investment, trade disruption, and geopolitical challenges. He argues that companies embedding stronger intelligence into decisions can compound an advantage over time. 19

That is a company strategy, not a proven industry-wide outcome. The practical test will be whether agents can deliver reliable results in high-stakes workflows while fitting companies’ approval processes, supplier relationships, and existing systems.

How the agents are meant to handle complex work

CTO and co-founder Odhran O’Donoghue says Magentic is developing systems that go beyond limited-context AI interactions: agents should be able to diagnose a problem, plan a fix, take action, and see work through across large volumes of multimodal enterprise information. 19

In procurement, that could mean linking structured records—such as spend and order data—with unstructured materials including contracts, invoices, technical documents, and correspondence. The important distinction is between an assistant that suggests a next step and an agentic system designed to carry a workflow forward. Magentic is pursuing the latter. 919

Security and deployment claims

For enterprises, the value of an AI agent depends on controls as much as automation. Magentic says it offers zero-data-retention agreements with major AI providers, flexibility to deploy in customers’ cloud environments, and isolated deployments in selected data regions. 2

Its earlier launch materials also said data is encrypted and cited SOC 2 Type II, ISO 27001, and GDPR compliance. 7 Prospective customers should independently validate the scope of these controls, integration permissions, human approvals, and auditability before allowing an agent to take actions in procurement or finance systems.

Where the $18M will go

Magentic says the Series A will accelerate its AI-agent roadmap, expand coverage across procurement and supply-chain workflows, and support longer-horizon research into complex industrial optimization. 1819

For manufacturers, the proposition is straightforward: use AI agents to connect the data, communication channels, and systems already involved in purchasing. The harder question—and the one that will determine adoption—is how well those agents perform safely and consistently when real supplier, contract, and production constraints collide.