For the last five years, Michael Manapat and I have taken long walks around Noe Valley talking about problems worth solving.
We kept circling back to a pattern we'd seen play out across Emergence's portfolio: the most successful founders bring lived experience to the problem they're tackling. Eric at Zoom had been VP of Engineering at Webex. Peter at Veeva had run Salesforce's platform. The founders who win aren't just smart. They've been inside the machine.
Michael had been head of ML at Stripe and CTO of Notion, where he helped usher the company into the AI era. He had a rare understanding of how to take messy, proprietary datasets and structure them so AI can actually reason over them in high-stakes environments.
As we brainstormed problem spaces, we kept returning to investing: an arena with a tremendous amount of proprietary data but massive barriers to making use of it. If AI is going to matter in finance, it has to reason over a firm's private history. Most tools don't clear that bar.
Today, we're thrilled to co-lead Rowspace's Series A alongside Sequoia, with $50M in total funding across both the seed and Series A. This partnership has been years in the making.
Drowning in Data, Starved for Structure
Global financial institutions manage well over $100 trillion in assets. The largest private equity firms and credit platforms have decades of proprietary deal flow, internal memos, covenant analyses, operating models, portfolio reviews, and post-mortems. In theory, this is a dream input layer for AI.
In practice, the data is fragmented across SharePoint folders, CRM systems, Snowflake tables, email threads, legacy file structures, and analyst-built spreadsheets that were never meant to scale. Institutional knowledge is distributed across people and systems in ways that are difficult to unify and nearly impossible to reason over programmatically.
The generative AI wave has triggered a surge of vendors promising to unlock this knowledge, layering LLMs over financial workflows to retrieve information and draft summaries. Some can even answer complex queries across documents. But retrieval is not reasoning. The real economic value sits in the gap between the two.
When models are pointed at messy, inconsistently structured internal data, they inherit that mess. Hallucinations increase. Edge cases break. Conflicting documents surface without context. In credit underwriting or portfolio monitoring, "mostly right" is not good enough.
That's the structural flaw in most first-generation AI products for finance: they skip the hardest step, which is transforming private data into something that can be reliably reasoned over.
The Hardest Step Is the Most Valuable One
The market already knows this. Firms are desperate to clean and standardize their own information. Knowledge process outsourcing firms like Acuity Knowledge Partners and Crisil generate close to nine figures in annual revenue from financial data structuring, reconciliation, and analytical support. The market is projected to reach $157.5 billion by 2030, with financial services driving a significant share of demand.
Today, that work is almost entirely labor. Once a report is delivered or a dataset is reconciled, the institutional knowledge embedded in that process dissipates back into static files.
Rowspace starts from a different premise: this transformation layer should not be a service. It should be infrastructure.
Rowspace starts with the foundation — unifying and structuring fragmented firm data — before layering intelligence on top. That means reconciling inconsistent schemas across systems, structuring historical memos into machine-readable formats, normalizing financial models, and mapping how a specific firm actually resolves conflicting information. Only once that foundation is in place do they layer search and analytical agents on top.
It's the only path to reliable, high-stakes AI.
In practice, this plays out in distinct and powerful ways. A PE firm evaluating a new deal can draw on decades of institutional knowledge to assess risks and opportunities. A growth investor making portfolio allocation decisions can act on what's true today, not numbers that will take weeks to reconcile. A credit investor can surface new opportunities matching their macro view while ensuring compliance at both the loan and portfolio level.
Firms managing hundreds of billions to nearly a trillion dollars in assets are already using Rowspace for exactly these workflows. They chose Rowspace because generic AI tools couldn't deliver the specificity and uncompromising accuracy their decisions require.
Public Intelligence Is Commoditized. Institutional Judgment Is Not.
Foundation models are getting good fast. Public data is everywhere. Surface-level insight is becoming cheap. But the need for smart judgment remains.
The partner who has seen five hundred deals recognizes patterns others miss. The credit investor who has lived through multiple cycles knows which covenant breaches are noise and which are signals. The growth investor knows how to interpret conflicting operating metrics because they've seen the second- and third-order effects play out before. That judgment is enormous, and it's trapped.
Rowspace's thesis is that AI in finance only creates advantage if it can reflect how a specific firm reasons: how it resolves conflicts, what it prioritizes, which historical precedents matter, and why. When that reasoning becomes structured and auditable, institutional memory starts to compound. A firm doesn't just store documents. It stores decisions and the logic behind them.
As Michael puts it: "Imagine a firm that never forgets. Where an experienced investor's workflows can be codified and multiplied. A first-year analyst can tap into decades of institutional knowledge, and judgment scales with the firm instead of being diluted."
Built by People Who've Lived It
I've known Michael for more than five years. We've explored dozens of ideas together. The through-line has always been his refusal to shortcut the foundational work.
Michael ran applied machine learning at Stripe across Radar, Capital, and Sigma, owning fraud, credit, and forecasting systems where reliability and precision were existential. He later served as CTO of Notion, where he built and scaled high-performance engineering teams while pushing into AI ahead of the market.
His co-founder Yibo Ling brings the complementary perspective of a two-time CFO who has synthesized fragmented systems to make real capital allocation decisions. He has lived the operational friction Rowspace is solving.
This is not a team experimenting at the edges of finance. They understand both the data infrastructure and the decision-making realities of capital markets. That combination is rare. Alfred Lin and I will both be joining the board, and we couldn't be more excited to work alongside them.
Why This Matters
At Emergence, we back founders who bring lived experience to big, enterprise-scale problems. The Rowspace team is the definition of that.
For AI in finance to work, firms need systems they trust with consequential decisions. Trust, in this context, is architecture: data never leaves the firm's infrastructure, every answer is traceable, conflicts and risks are exposed early, and reasoning is grounded in the firm's own record rather than an abstract market average. Knowledge doesn't need to reset every time the org chart changes.
Public intelligence is commoditized. Private reasoning is not. After years of discussing what it would take to build something durable in this space, we're excited to make our partnership with Rowspace official.
Welcome to the Emergence family, Michael, Yibo, and the entire Rowspace team.
