LevaData

PLEASANTON, CALIFORNIA - FEATURE

Here is a fact about manufacturing that sounds made up but isn't: the single largest line on a company's cost sheet is very often the one it understands the least. Direct materials - the chips, connectors, boards, cells and cases that go into a physical product - can eat half of the cost of goods sold. And for years, the software that helped companies buy them was, roughly, a spreadsheet, a buyer, and a phone call. Indirect spend, the pens and printer paper and travel bookings, got the shiny procurement suites. The stuff that actually determined whether your product shipped on time and at margin got a VLOOKUP.

LevaData was founded in 2015 to sit precisely in that gap. Its pitch is narrow and, once you hear it, hard to unhear: build one AI-powered platform that treats direct-materials sourcing as its own discipline, feed it the messy reality of your bill of materials, and point it at real-time market data covering more than a billion parts from tens of thousands of suppliers. The company calls this "cognitive supply management." What it means in practice is that the software is constantly re-scoring your parts - for cost, for risk, for lead time, for tariff exposure - and surfacing the moves a buyer might otherwise find three weeks too late.

That framing matters because of a structural quirk of the problem. Cutting cost and building resilience usually feel like opposing goals. The cheapest supplier is often the riskiest; the safest dual-source arrangement costs more. LevaData's argument is that the trade-off only looks binary when you can't see the whole board. Give a team live visibility into part risk, alternate parts, supplier scorecards and location impact all at once, and the two objectives stop competing quite so much. You are no longer choosing between cheap and safe in the dark. You are choosing with the lights on.

The founder, Rajesh Kalidindi, built the company around a thesis that supply management should be a source of edge rather than a cost center to be minimized. It's a nice line, and it's also the kind of line that only pays off if the boring plumbing underneath actually works. And the boring plumbing here is genuinely hard. To reason usefully over a billion parts, you first have to reconcile a billion parts - matching a customer's internal part numbers to manufacturer part numbers to market feeds, cleaning the duplicates, resolving the aliases. Most of the "AI in procurement" story is really a data-integration story wearing a nicer jacket. LevaData spent years on the jacket's lining.

What sits on top is a set of tools with deliberately literal names. There is Part Insights, for part-level analytics. Spend and Savings Intelligence, for finding the money. Supplier Scorecard, for ranking who actually delivers. Risk, Tariff and Location Insights, for the things that ruin a quarter. And Smart RFX, which folds the entire quote-bid-award negotiation loop - the RFQ, the responses, the comparison, the award - into a single screen backed by market data. The naming is not accidental. Procurement buyers are not looking for whimsy. They are looking for the part that is about to go end-of-life and the three qualified alternates that can replace it before the line stops.

The money has followed the thesis, if not at hyper-growth velocity. LevaData raised early venture funding from Tola Capital, added a reported $12M Series B in 2018, and then, in September 2021, closed a $47M Series C led by Banneker Partners with Tola participating - bringing total funding to roughly $65 million. That 2021 timing is worth pausing on. It landed in the middle of a global supply-chain nervous breakdown, when semiconductor shortages and shipping chaos had turned "direct materials" from a procurement-team concern into a CEO-and-board concern. A company selling instruments for exactly that problem was suddenly selling into a market that had discovered, painfully, why the instruments mattered.

Investors described the appeal in plain terms. Direct materials, as Tola's Aaron Fleishman put it, "impact product brand, margins, and ability to meet demand." That is the whole case in a sentence: this is not back-office cost control, it is the difference between having product to sell and not. Banneker's Stephen Davis framed LevaData as positioned to help companies "maintain resiliency while navigating supply complexities" - venture-speak for "the world just proved this is load-bearing."

In November 2022, the board handed day-to-day leadership to Keith Hartley, naming the supply-chain software veteran as chief executive. Hartley's résumé is a tour of the category: manufacturing sales leadership at Blue Yonder, senior roles at Oracle, most recently SVP of Sales at Ivalua, plus an MBA from London Business School. He talks publicly about applying first-principles thinking and design thinking to procurement, and about leading with what he calls servant leadership - the sort of language that can read as filler until you remember that enterprise supply-chain software is sold, painfully, one skeptical buyer at a time. Hiring a career seller of exactly this software to run a company that makes exactly this software is not a subtle signal about the next chapter's priorities.

Who actually uses this? LevaData aims at both large, complex enterprises and small-to-mid-sized OEMs - anyone building physical things with a bill of materials complicated enough to hide problems in. Publicly referenced customers span recognizable hardware names, and the company sits inside a partner ecosystem alongside supply-chain planning players and consultancies, which is how enterprise software of this kind tends to reach the enterprise: not alone, but wired into the systems and advisors a manufacturer already trusts.

The honest read on LevaData is that it picked a problem most people would rather not touch. Direct-materials sourcing is data-heavy, unglamorous, and genuinely difficult to get right. There is no consumer app to demo at a party, no viral loop. There is instead a billion messy parts, a tariff schedule that changes under you, and a buyer who will not care about your AI unless it hands them a better alternate part than they could find themselves. LevaData's bet, a decade in, is that this unglamorous corner is exactly where the value hides - and that a company patient enough to clean the data can turn procurement from a place where margin quietly leaks into a place where it is defended. Whether that bet fully pays off is still being written. But the choice of where to plant the flag looks, in hindsight, less like a gap in the market and more like a market that hadn't yet noticed the gap.