Arm’s finance and procurement teams cut contract review time 50% with WisdomAI

Overview

Arm is a semiconductor and software company that designs high-performance chip architecture, CPU cores, and software development kits. Arm’s architecture is found inside the majority of the world’s smartphone chips, and the company has ridden a wave of AI infrastructure demand to three consecutive years of over 20% revenue growth, with annual revenue now approaching $5 billion.

Founded in 1990 and headquartered in Cambridge, England, Arm is publicly listed on the Nasdaq stock exchange, with more than 8,300 employees and a market capitalization in excess of $250 billion as of August 2026.

As Arm scaled, its procurement and accounts payable functions began rethinking how spend and contract data were analyzed, and whether that data could support cash flow, cost, and risk decisions at the executive level.

Challenge: Every question becomes another ticket

A few years back, Arm’s procurement team ran into a problem familiar to any data-driven organization operating at scale: the information the team needed to answer questions about supplier and vendor contracts usually existed somewhere, but exactly where was anyone’s guess.

For example, spend data was split across an ERP system, credit card statements, supplier documentation, and Amazon receipts. To pull together something as basic as total spend, the team had to reconcile inconsistencies between systems by hand and then add in business context that didn’t live in any database.

“A lot of the knowledge about how procurement operated at Arm wasn’t really in the systems,” says Thomas Smith, Procurement Transformation Lead at Arm. “It was in people’s heads.” 

Further complicating matters were reams of unstructured contract data. While metadata could confirm whether a termination clause existed, it couldn’t confirm what it said, whether termination carried penalties, or under what conditions it applied. These were the kind of details Smith’s team needed every quarter for contract reviews, and no number of additional mandatory metadata fields could fully capture them.

The problem extended beyond procurement’s own reporting. Source-to-pay data and contract documents lived in separate tools, disconnected from the payment data finance teams needed downstream to hold suppliers to negotiated terms. This setup made it tricky to trace a clear line from a contract’s terms to its financial impact. While legacy dashboards could show what a metric was doing, they couldn’t show why or recommend what to do about it. 

The lack of confidence reached all the way up to executive reporting. Teams weren’t fully confident in the numbers rolling up to leadership in areas like liabilities and accounts payable. This made it harder to prove the value of procurement’s own initiatives, such as savings, productivity, and cost avoidance, especially anywhere that impacted cash flow.

The cost also showed up in how Smith’s team spent its time. Every question that didn’t map onto an existing dashboard became a standalone request. Every follow-up, whether a missing column, a different filter, or a different date range, became another one.

“Every follow-up effectively is another reporting request. So that’s another ticket. That’s more hours of time.” 

—Thomas Smith, Procurement Transformation Lead, Arm

Solution: Teaching WisdomAI how Arm operates

To solve these problems, Arm turned to WisdomAI, which unified on-demand insights across structured and unstructured spend, supplier, and contract data wherever it lived, letting the procurement team query all of it directly from one place. With WisdomAI, the contract language that would never fit cleanly on any dashboard became something the team could simply ask about in plain language, with no SQL required. That visibility surfaced ideas the team simply hadn’t had before, like where supplier terms could be tightened or renegotiated.

The team had a working proof of concept running in about three weeks, much faster than a typical enterprise buildout. The initial results were immediately noticeable.

“We got some really good answers. We were really impressed.”

—Thomas Smith, Procurement Transformation Lead, Arm

But Smith and his team quickly learned that connecting the data was only the beginning. Analytics agents need context to work accurately. The real work was teaching WisdomAI Arm’s context — how the company’s procurement team operates: its workflows, business rules, semantics, and how it defines and calculates core metrics.

WisdomAI’s Adaptive Context Engine gave Arm’s team specific ways to build and maintain that understanding:

• Business context that eliminates ambiguity quickly: When an analytics agent mistook a company that was both a supplier and the name of a product category Arm tracked separately, one line of context — “Company A is a supplier of X” — resolved it for good. The fix persisted across every related question moving forward.

• Continuous testing and refinement: The procurement team tested as many as a hundred variations of a single question against a dataset to validate that WisdomAI returned the correct interpretation. They also reviewed the SQL behind each answer and investigated every thumbs-down signal. As a result, day-to-day usage contributed to greater accuracy in a continuous feedback loop.

• Built-in evaluation tools that catch conflicting context: As adoption grew, Arm introduced a formal review step before new context went live. In conjunction with WisdomAI’s context evaluation feature that flags contradictions between existing rules, the team was able to surface internal playbooks that disagreed with each other and make appropriate fixes.

“We thought more context was going to improve answers,” Smith says. “What we’ve actually found is the opposite. Less context, more very specific, very refined context.” 

That same context foundation let Arm connect the full financial thread behind a purchase, from NDA and MSA through purchase requisition, purchase order, invoice, and payment. This gave finance and procurement a single, traceable line from contract terms to cash impact, something no dashboard had been able to show end-to-end.

Arm started with a small rollout, deploying WisdomAI to roughly 40 users across procurement instead of pushing it company-wide. Within that pilot, the team built reusable structures called “category packs,” which included predefined views of spend and supplier performance for a category that any team member could access just by asking for it.

With a sturdy context foundation in place, Arm built agentic monitoring on top of it, letting agents watch for changes in operational metrics like cycle time, flagging them automatically. Agents also track KPI performance against annual targets, making it easier to catch likely misses before they happen. “It’s self-governing, which is wonderful,” Smith says.

“You’re always going to end up with conflicts. We put in a load of our standards and playbooks, and then you find that one of our playbooks is totally conflicting with another one. Like, instantly. Just through doing this, we’re actually solving an operational business problem, too.” 

—Thomas Smith, Procurement Transformation Lead, Arm

Results: Reclaimed time, trusted numbers, and faster decisions

The biggest win for Arm came during the company’s quarterly contract review process. Understanding liabilities, payment milestones, and commercial terms across hundreds of contracts had been a grueling, resource-intensive manual process that took hundreds of hours each quarter. With WisdomAI handling contract analysis, Smith estimates the team cut that time by 50–100 hours, even with what he describes as an early stage implementation.

“Wisdom halved that,” Smith says. “We still had to do manual work, but we probably knocked off 50-plus hours a quarter, maybe even close to 100, straight off the bat.” 

That same caliber of contract intelligence extended beyond routine reporting. For example, when Smith needed to review a set of SOX (Sarbanes–Oxley) disclosures, existing metadata wasn’t detailed enough to understand the contracts involved. WisdomAI solved this problem, pulling directly from the contract language itself while also proactively suggesting areas for further discovery.

WisdomAI’s accuracy held up even without the additional context Arm’s procurement department continued to provide. In a separate analysis of non-cancelable contract commitments — work that had previously taken close to two person-months to complete by hand — the initial WisdomAI output came in at 98.5% accuracy against the organization’s own benchmark. Full accuracy is expected as more context is added. 

With clearer, traceable connections between contracts, transactions, and payments, Arm’s finance and procurement teams reported:

• 50% reduction in time-to-decision: Teams moved faster from a KPI to an explanation to a recommended action instead of staring at a static number on a dashboard.

• 1.5x increase in Net 60 payment term adherence: Better visibility into vendor terms and payment timing let teams more consistently hold suppliers to negotiated terms, directly supporting cash conservation.

• 19% in cost savings: The impact was enough to help self-fund the procurement organization’s own annual operations.

As procurement got more comfortable asking WisdomAI questions directly, the volume of reporting requests to the data team dropped considerably:

• Significantly fewer reporting tickets: With roughly 40 users able to query data conversationally, fewer questions end up as formal requests to Smith’s team.

• Faster strategic work: One category manager told Smith that building a category plan, which was previously a multi-step reporting exercise, took roughly half the time once he could pull spend insights directly through WisdomAI.

• Deeper self-service exploration: Smith, who can view query activity across his team, describes users going “down rabbit holes” of follow-up questions in a way static dashboards simply don’t allow.

WisdomAI’s agentic capabilities have also delivered, surfacing findings the team wouldn’t have caught on its own. For example, when cycle time monitoring flagged a delay in Arm’s PR-to-PO process, the agent traced it back to a small increase in approvals on purchase orders between $10,000 and $50,000. This prompted a conversation about whether that approval threshold made sense anymore. 

“That’s just awesome because that’s driving actual business and strategic conversations just in and of itself,” Smith says.

With fewer tickets to wade through, Smith’s team redirected its time toward more important work.

“That’s freeing up my time to improve, and the team’s time to improve, data quality, which is still so important, and governance, which is king, rather than just producing reports.”

—Thomas Smith, Procurement Transformation Lead, Arm

Looking ahead: Context as an operating discipline

For Smith, the biggest benefit WisdomAI delivers is a shift in what his team spends its time on. Context management and governance now account for roughly one-quarter of the team’s work, and that share is still growing. Work that used to involve building dashboards and pulling reports increasingly means reviewing query behavior, examining the SQL behind an answer, and proposing refinements to the context layer itself.

“Context management and context governance are increasingly becoming a part of my role,” Smith explains. “That’s a totally different problem from just building dashboards, and you’re finding that all of our roles are starting to reflect that.” 

This shift is changing what Smith looks for in candidates when he hires. While technical skills like SQL still matter, they’re no longer the top qualification.

“I’d say the most important thing is going to be curiosity,” he says. “You need somebody that’s interested in how organizations, how processes work and just wants to ask questions about them and sort of be like, ‘How can this be better?’”

As Arm adopts more AI tools across the business, he expects the context his team has built to become foundational infrastructure.

“Everybody knows this. Individual data engineers know you have to clean this data this way,” Smith says. “Getting that all in one place, in a dictionary, and managing it — that’s actually a wonderful and very valuable thing that I don’t think we’re underestimating.”

“It’s a massive shift. It’s less time and less resources producing information. It’s more time understanding the information.” 

—Thomas Smith, Procurement Transformation Lead, Arm