Fragment, a San Francisco startup building AI agents for corporate purchasing, launched publicly on Oct. 28 with $56 million in funding, co-founder Pranav Mulgund said in a company blog post.
The total covers two rounds.
8VC led a Series A, and General Catalyst and A* co-led an earlier seed round, Tech Funding News reported. Angel investors include Roelof Botha, Nikesh Arora, Jeff Wilke, Eric Wu and Dennis Cinelli, according to Fragment. The law firm Goodwin said it advised the company on the Series A.
Fragment is starting with direct procurement, the buying of the materials and parts that go into a manufacturer’s products. Mulgund wrote that it picked that area on purpose, because “complex global supply chains and fragmented systems have made automation particularly difficult” there.
The work it targets
The problem is exceptions. When an order changes, a delivery arrives in pieces or an invoice doesn’t match the agreed price, someone has to dig through several systems to work out what happened and fix it, Mulgund wrote. He said that work “happens constantly and quietly across global businesses,” and it’s grown harder with shifting tariffs and supplier disruptions.
Large companies already run purchasing software such as SAP, Ariba and Coupa, Mulgund noted. His argument is that those systems record the work but still leave people to clean up when something goes wrong.
Fragment’s answer is what it calls an Autonomous Context Engine, or ACE. The company says it connects a customer’s data, systems, policies and institutional knowledge into “an operational blueprint of how the business works.” Fragment’s agents then use that to investigate a discrepancy, decide the next step and carry out an approved fix.
The agents work inside the systems a company already uses. “Enterprises have spent decades building their technology infrastructure and they shouldn’t have to replace it to benefit from AI,” Mulgund wrote.
Early customers
Fragment names Graybar, a distributor of electrical and communications equipment, and CNX, an energy company, as customers. Graybar processes about 3 million invoices a year, and 80% of them need some kind of resolution, according to Fragment. That figure comes from the company.
Tech Funding News reported that about 20% of transactional procurement is still done by hand. The outlet also said the founders see scattered data, not weak AI, as the main obstacle to using agents in big organizations. Procurement staff often pull information from many systems and fall back on informal rules to get things done, it reported.
Mulgund worked as a physics researcher and led part of Zoom’s product team before starting Fragment, according to Tech Funding News. He co-founded the company with Naveen Kashyap. Mulgund wrote that manufacturing is personal to him, because his parents met in a factory.
He also reached back to 1887, when railroad executive Marshall Kirkman wrote a manual on how big companies should manage purchasing and inventory. Many of those same jobs still exist nearly 140 years later, Mulgund wrote.
Fragment says it’ll use the money to develop the ACE further and expand across source-to-pay, the full cycle from choosing a supplier to paying the bill. “Procurement is where we’re proving what’s possible, but our ambition extends much further,” Mulgund wrote.
Source: Fragment, Pranav Mulgund, “We’ve Raised $56 Million to Make Enterprise Software Do the Work”
Additional sources: Tech Funding News, Sofia Chesnokova, “Fragment raises $56M from 8VC, General Catalyst and A* to put AI agents on procurement exceptions”; Goodwin, “Goodwin Advises Fragment Data Technologies on its $56 Million Series A Financing”
