Magentic Raises $18M Series A to Build an AI Workforce for Industrial Operations

Magentic has raised $18 million in Series A funding to expand its AI agents across procurement and supply chain operations, targeting some of the most complex workflows inside large industrial companies.

The round was led by Felicis, with existing investors Sequoia Capital and The Westly Group also participating. It comes roughly a year after the London- and New York-based company launched with a $5.5 million seed round, bringing its reported total funding to $23.5 million.

Founded by CEO Robin Van Aeken, formerly of McKinsey, and CTO Odhran O’Donoghue, an OpenAI alumnus, Magentic is betting that the next stage of enterprise AI will move beyond assistants that answer questions or generate content. Instead, its agents are designed to carry out operational work across the systems manufacturers already use.

That puts the company at the intersection of two major trends: the rapid expansion of agentic AI and a capital-intensive buildout of the physical infrastructure required to support the broader AI economy.

Why Industrial Procurement Is Emerging as an AI Target

Much of the generative AI boom has focused on software development, customer service, marketing, and other predominantly digital workflows. Magentic is going after a less visible but economically significant target: the processes that determine what companies buy, from whom, at what price, and under which contractual terms.

Those processes become particularly difficult inside global manufacturers. Procurement data can be scattered across enterprise resource planning (ERP) platforms, supplier agreements, spreadsheets, invoices, email systems, and databases accumulated through years of acquisitions and technology changes.

At the same time, pressure on procurement organizations is increasing. The Hackett Group projects that procurement workloads will grow by 8% in 2026 even as headcount and operating budgets decline, creating an incentive to automate more of the work without simply expanding teams.

There is also a much larger infrastructure cycle underway. Goldman Sachs estimates a baseline of approximately $7.6 trillion in cumulative AI-related capital expenditure between 2026 and 2031, spanning compute, data centers, and power infrastructure. That spending ultimately translates into an enormous volume of equipment, materials, construction, supplier relationships, and contracts that need to be managed.

Magentic’s thesis is that AI can increasingly participate in those decisions rather than merely helping humans analyze them.

Moving Beyond the AI Copilot

Magentic describes its agents as “digital workers,” but the more important distinction is the degree of autonomy the platform is attempting to provide.

Rather than offering employees another dashboard or conversational interface, Magentic is designed to sit on top of existing enterprise infrastructure. Its architecture includes data ingestion, a harmonized data layer, an agentic engine, procurement-specific knowledge, real-time monitoring, feedback loops, and end-to-end execution.

The system can ingest fragmented information from ERPs, spreadsheets, PDFs, contracts, and other sources before turning it into a structure that AI agents can work with. Magentic also says the platform can select different underlying AI models depending on the task, rather than relying on a single large language model.

That matters because procurement involves far more than retrieving information.

An agent might identify that the same component is being purchased at different prices across factories, find overlapping suppliers that could be consolidated, detect purchases being made outside negotiated contracts, or discover rebates and contractual discounts that were never collected.

Magentic’s goal is for the agent to go further and help complete the resulting workflow.

Its digital workers can review spend, examine contracts, calculate potential savings, prepare supplier communications, reconcile orders against invoices, and assemble the evidence required to recover overpayments. The company says its agents can operate across spend, sourcing, contracts, and invoices rather than being limited to one isolated procurement task.

AI Agents Built Around Messy Enterprise Data

The underlying data problem could prove just as important as the AI models themselves.

Large manufacturers rarely have pristine databases waiting for an AI system. Supplier names may differ between divisions, contracts may exist as PDFs, prices can be recorded differently across ERP instances, and decades of acquisitions can leave companies with numerous overlapping systems.

Traditional enterprise transformation projects often attempt to clean and standardize that data before deploying new applications.

Magentic is taking a different approach: allowing AI agents to work across the fragmented environment that already exists and progressively organize the information required to complete a task.

The platform is also designed to communicate through tools employees already use, including Microsoft Teams and email, while operating across internal enterprise systems. This is intended to make the agent less like a standalone software product and more like another operational participant inside an existing process.

Magentic says one customer is already processing more than 1.2 million orders through its digital workers, while another deployment identified $4 million in savings. Across its Global 500 customer base, the company reports savings of roughly 2% to 5%, alongside an average 60% improvement in data quality. These are company-reported figures rather than independently audited performance metrics.

Keeping Humans in the Procurement Loop

Greater autonomy also creates a problem that is particularly important in procurement: an incorrect AI decision can have immediate financial or operational consequences.

Sending the wrong supplier communication is considerably different from generating an inaccurate paragraph in a chatbot.

Magentic has therefore built human review points into its agent workflows. In one example described by the company, an agent can scan supplier contracts and flag unusual payment terms, but a procurement executive retains responsibility for deciding whether a negotiation should actually take place.

The company has also described a multi-agent review mechanism in which agents can check the work produced by other agents, alongside logs and evidence trails intended to show how decisions were reached.

This human-in-the-loop model is likely to remain important as enterprise agents move from recommending actions to executing them.

Security Becomes Part of the Agent Architecture

Giving an AI system access to contracts, supplier relationships, pricing information, invoices, and internal ERP systems also raises a different set of enterprise requirements.

Magentic says it is SOC 2 Type II and ISO 27001 certified, compliant with the European Union’s General Data Protection Regulation (GDPR), and aligned with the EU AI Act.

The platform includes controlled interactions with systems of record and human checkpoints for important actions. Magentic also says customer data is not used to train shared or public models.

These safeguards become more consequential as agents gain the ability to act rather than simply retrieve or summarize information.

Building an AI Workforce for the Physical Economy

Magentic plans to use the Series A to expand the range of procurement and supply chain workflows its agents can manage while investing further in the underlying AI research.

One technical challenge is context.

Industrial optimization problems can involve years of purchasing data, thousands of supplier contracts, pricing histories, inventory information, invoices, technical documents, and communication records. Magentic says it is developing agents capable of diagnosing problems, planning solutions, taking actions, and following those actions through across terabytes of multimodal information.

“Bringing frontier AI to the physical world requires pushing beyond AI systems with limited context windows,” said O’Donoghue. “We’re building AI that can diagnose problems, plan the fixes, take action, and see the work through across terabytes of multimodal data at once.”

The broader shift is from AI as a tool used during a workflow toward AI as software capable of owning increasingly large portions of that workflow.

Procurement offers an interesting test case because its outcomes are measurable. A system either catches an incorrect price, prevents off-contract spending, identifies an overlooked rebate, reconciles an invoice, or it does not.

For Magentic, the $18 million Series A is therefore about more than expanding another enterprise AI product. The company is betting that some of the most valuable AI agents will ultimately operate far away from the chatbot interface—inside the complicated procurement systems that determine how the physical economy actually gets built.