Saumil Annegiri

The credit market is a paper problem.

A private credit deal is largely a stack of PDFs. A credit agreement runs a few hundred pages. A term sheet, a covenant package, an intercreditor arrangement, an amendment, a waiver. Somebody has to read it. Somebody has to compare it against the last one. Somebody has to notice that a definition moved three sentences and now the borrower can incur half a turn more leverage. That somebody is usually a junior analyst at 10:47 p.m. This is the market that Saumil Annegiri, Co-Founder and Co-CEO of CredCore, has decided to automate.

Annegiri arrived at that decision through the front door. Before CredCore, he was Managing Director for Technology and Digital Infrastructure at Guggenheim Partners, where he built the digital infrastructure business - structuring equity and debt investments in data centers, fiber, and infrastructure software. Before Guggenheim he was at Deutsche Bank, working on communications infrastructure, advisory, and structured products. Before that: SunGard Availability Services on the portfolio side, and, earlier, work with the U.S. Department of Defense on artificial intelligence and supply chain. His stated career count is more than $50 billion in transactions executed. He has been the customer he is now selling to.

In 2022 he co-founded CredCore with Karthik Nandyal. In February 2025 the company announced a $16 million Series A led by Avataar Ventures, with Inspired Capital, Fitch Group, and BellTower Partners participating alongside senior executives in asset management and financial services. CredCore describes itself as an AI-native software platform purpose-built for institutional credit, a phrase that matters because most fintech platforms are neither AI-native nor purpose-built for anything except a Series A deck. CredCore says its models have been trained on roughly $5 trillion worth of debt data, and that the platform is in production at asset managers overseeing more than $650 billion in AUM.

The pitch, stripped down: read the documents faster, extract the clauses, compare the deals, track the covenants, generate the audit-ready reports, and hand analysts something they can defend to a compliance officer. Do it in hours instead of days. Do not hallucinate. Annegiri's answer to the hallucination problem, in his telling, is domain-specialist verification layered over the models. The AI does the first pass; a person who has actually negotiated a credit agreement checks it. This is unglamorous in a way that is probably load-bearing.

The two-CEO arrangement deserves a note. Startups this size usually pick a CEO and call the other founder president, CTO, or COO. CredCore did not. Annegiri and Nandyal both hold the CEO title. It is an uncommon choice at Series A and it forces the two of them to work in a way that most first-time co-founders never learn. It is also legible if you think about what the company sells: institutional credit is a domain where nobody trusts a generalist. Splitting the top of the org chart is a way of saying that both the technology side and the credit side are load-bearing, and neither reports to the other.

Why this market, why now.

Enterprise credit is often described as an underserved software market, which is the polite way of saying it has been left alone. Assets under management have grown. Private credit funds have raised record vehicles. Direct lending is now a category that institutional allocators discuss out loud. The workflow that produces those deals, however, has aged like paper. Excel models, PDFs, shared drives, email threads with "v27 FINAL clean" in the filename. If you have ever seen a data room for a credit deal you understand the joke. Annegiri's insight is not that AI will replace credit analysts. It is that AI can eat the ninety percent of the analyst's day that is document handling, so that the ten percent that is judgment gets more room.

CredCore's product surface, in the language of its website and press releases, includes credit agreement extraction, clause collaboration and redline tracking, covenant monitoring, obligation checklists, deal comparison, and audit-ready reporting. In plain English: it reads the paperwork, tells you what is in it, tells you what changed since last time, and produces the documentation your regulator will want when they ask. It sits between the analyst and the deal document, and it does the same job for the pre-deal read as for the post-deal monitoring.

The investor list is a read of its own. Fitch Group's participation is a signal about ratings and analytics adjacency. Inspired Capital brings early-stage credibility with financial services operators. Avataar Ventures leads with a growth-stage vertical AI thesis. The senior asset-management and financial-services executives participating alongside are, in effect, the customer base doing due diligence on the product by writing checks. Founders sometimes complain that strategic capital comes with strings. In credit, it comes with distribution.

The résumé, read sideways.

The interesting thing about Annegiri's background is not that he has finance credentials. Half of Wall Street has finance credentials. It is the shape of the credentials. A bachelor's from the University of Pune. A master's from the University of Louisiana at Lafayette. An MBA from Wharton. Time at the Department of Defense on AI and supply chain problems. Portfolio work at SunGard. Structured products at Deutsche Bank. A Managing Director seat at Guggenheim where he built a digital infrastructure business from scratch. Each of those is preparation for a different company. Stacked together they are preparation for exactly this one - a fintech that has to be credible with a credit committee, a compliance officer, a model risk team, and a machine learning engineer at the same time.

He has also spent enough of his career financing the infrastructure that AI runs on - data centers, fiber, edge - that the decision to leave and build an AI company reads less as a bandwagon than as a completion. The person financing the picks and shovels eventually decides to swing one.

The near future.

The company said the Series A would go to expanding AI capabilities, growing the team, and supporting a broader range of credit market participants and deal types. That is the standard Series A press-release commitment. The specific version, in Annegiri's case, will probably show up as more coverage of asset classes: direct lending today, private ABS and structured credit tomorrow, syndicated broadly held credit further out. Each of those has its own document conventions and its own audience of analysts. Vertical AI in credit is not one product; it is a series of vertical products that share a data foundation. That is a slower expansion than pure horizontal SaaS. It is also harder to displace once it is in.

The competitive question is the one every enterprise AI company faces: does the incumbent build this or buy it? Rating agencies have analytics arms. Bloomberg has terminals. Broadridge has back-office. Ratings, terminals, and back-office are all adjacent, none of them are the same product, and each of them is famously slow to ship. Annegiri's advantage is speed measured against those clocks, not against a peer startup. This is the kind of race that founders with two decades of domain time tend to win.

The market is $5 trillion. The workflow is a PDF.

Annegiri's stated goal is to compress the debt deal lifecycle - pre-deal read, deal execution, post-deal monitoring - from days to hours. If it works, CredCore becomes the software layer of a market that has never really had one. If it does not, it becomes another AI wrapper. The interesting bet is that his career has trained him to know the difference.