From Goldman To AI: How Rishi Bali Is Building Wall Street’s Transformation Layer With OPCO.AI

Rishi Bali, Founder & CEO of OPCO.AI, outside of his office on Wall Street

RISHI BALI

Finance runs on spreadsheets. OPCO.AI thinks that’s a $1 trillion risk. Wall Street’s next alpha isn’t in trading — it’s in AI infrastructure. Across global capital markets, the infrastructure gap between trading and risk operations has become as critical as the strategies themselves.

In an expanding credit cycle, a missed covenant, a delayed NAV, or a reconciliation error can cascade across portfolios. Private markets still run on PDFs and manual processes — fragile foundations for a $13 trillion industry..

One firm betting that alpha now depends as much on the pipes as on the trade is OPCO.AI—founded by Goldman veteran Rishi Bali. OPCO.AI describes itself as Wall Street’s transformation layer — a builder that fuses risk, operations, and technology into a single, AI-powered operating core. After three decades building the machinery of modern finance — from Goldman’s credit desks and derivatives prime brokerage to multi-strategy hedge funds and private platforms — Bali is betting the edge in the next era won’t come from squeezing another basis point out of trades, but from sealing the basis points that leak out of broken workflows.

Where the Money Scaled — And the Infrastructure Didn’t

Private Equity & Private Credit. The buyout playbook thrived on cheap leverage. Then rates spiked to decade highs. Banks, constrained by post-GFC rules, stepped back from riskier lending, and private equity stepped in. Private credit — once niche — is now one of finance’s fastest-growing corners, fueled by deep insurance capital (think Apollo/Athene, KKR/Global Atlantic). But while capital has institutionalized, the operating infrastructure hasn’t. Covenants and borrowing bases live in PDFs. Amendments move through email. Risk engines are patchy; finance systems creak; investor reporting lags. In a benign cycle, those gaps stay hidden. Under stress, they become systemic weaknesses.

Growth-stage VC and crossover funds increasingly resemble PE in their structures and scale — and they too face rising expectations from institutional LPs for faster reporting, stronger controls, and more resilient infrastructure. Industry outlooks peg private markets at roughly $13 trillion today on a path toward $20 trillion by 2030, and fundraising in private credit has concentrated among experienced managers in the higher-rate era — a winner-take-most dynamic that raises the bar for operating maturity. The more immediate risk is toxic operations, not toxic assets—covenant data trapped in PDFs can be more dangerous than any single borrower.

The systemic-risk debate here is well covered elsewhere (see recent WSJ interviews with Jamie Dimon and Sixth Street’s Alan Waxman, and FT coverage on opacity/retail exposure); this piece focuses on the operating spine rather than relitigating that debate.

Counterpoint (brief): Not all agree private credit is systemic. FT has argued that fund lockups, hard redemption caps, and diversified lender bases blunt contagion risk. The operating burden still rises either way — especially on covenant data, valuations, and investor transparency.

Hedge Funds. The two-and-twenty model faces a new math. Performance pressure in front; legacy systems in back. Many funds still run on 20-year-old risk/accounting stacks. Shadow accounting doubles effort; no-shadow outsourcing introduces fresh fragility. As complexity rises — from structured credit to CLO warehousing — manual reconciliations become existential. Allocators and regulators want near real-time views. Markets deliver macro, tariff, and geopolitical shocks on their own schedule. The result: sophistication at the strategy layer riding on creaky pipes underneath. Meanwhile, industry AUM has pushed to record territory — north of ~$4.5 trillion (HFR) — intensifying the need for industrial-grade controls and reporting. Recent fundraising barometers are widely covered (Bloomberg on Jain Global’s $5.3B debut; Bloomberg on D.E. Shaw’s human-run fund), underscoring a winner-take-most dynamic and closer scrutiny on infrastructure. In tight-spread markets, operational accuracy matters as much as strategy—the speed of reconciliation and provable lineage can outweigh another trading signal.

Family Offices. Founder liquidity created a wave of family offices — and a governance gap. Too many run on ad hoc spreadsheets, basic custodial statements, or the wrong jurisdictional setups. The tools exist (Addepar/Mirador), but tools without institutional discipline don’t create controls. Moving from concentrated founder wealth to institutional portfolios requires operating rigor — policies, risk frameworks, and consolidated reporting — or concentration and operational risks compound quietly. Estimates for the number of family offices range widely — from ~8,000 to well over 10,000 globally — underscoring both growth and opacity in this segment. The reputational risk isn’t performance; it’s process opacity—allocators notice sloppy dashboards and inconsistent closes.

Banks. Even with regulators watching, legacy cores and offshored processes have left surveillance, compliance, and trade-monitoring underinvested. For investment banks and capital markets players alike, financial governance and infrastructure modernization have become survival issues. Fines, model risk, and AI governance expectations have made “now or never” modernization unavoidable. Customers expect near real-time; fintechs set the bar; cyber risk expands with every integration. The U.S. shift to T+1 settlement compressed operational timelines further. Recent actions illustrate the cost of lagging controls: in 2024, U.S. regulators levied roughly $250 million (OCC) and $98.2 million (Federal Reserve) in penalties tied to surveillance deficiencies at a large U.S. bank; the CFTC added $200 million (with up to $100 million credit for overlapping issues). The choice is stark: pay later in penalties and reputation, or invest now in resilient operating backbones. Being “regulated” isn’t the same as being resilient; in 2025, model/AI governance is the new capital ratio.

Wall Street’s New Operating Paradigm: OPCO.AI’s Bet On AI

Bali’s conviction is that AI isn’t a bolt-on tool; it’s a new operating paradigm. OPCO.AI positions itself as building an Operations Intelligence Layer — a governed data spine that fuses risk, operations, and technology with explainable automation, clear lineage, and human-in-the-loop AI copilots.

His vantage point was earned in the trenches: Goldman Sachs Rates & Credit in the late 1990s, helping stand up the firm’s credit infrastructure as the asset class grew from a sidecar to a core engine; launching derivatives prime brokerage to support multi-asset hedge funds; then senior roles at D.B. Zwirn, UBS Dillon Read, Alphadyne, and Citigroup Tribeca Global Management. He also built a Made-in-USA technical apparel brand with institutional capital — a founder’s lesson in execution and design that now shows up in OPCO.AI’s doctrine: design is a control.

The OPCO.AI Doctrine

OPCO.AI is a builders’ shop, not a slide factory. Senior operators collapse risk, ops, and tech into a single source of truth, run pre‑mortems to contain blast radius, and treat design as a control. The team executes fast, hands the capability back, and measures success in hard outcomes: fewer exceptions, faster closes, cleaner audits, and basis points saved.

Institutional Lineage: Goldman Sachs Roots, Buy-Side Scale

At Goldman, Bali watched credit evolve from an offshoot of rates into a full asset class — investment grade, high yield, CDO/CLO warehousing, credit reinsurance — and helped architect the operating controls those desks needed to scale. Extending prime brokerage beyond equities into credit, rates, and FX required entirely new models of risk and reconciliation. Today, as private markets take on the scale and scrutiny once reserved for public markets, the operating burden has multiplied. OPCO.AI is his vehicle for applying that experience at buy-side scale.

How OPCO.AI Works: The Target Operating Model

Every engagement starts with a one‑line brief — the sharpest statement of pain (e.g., “Covenant monitoring is broken,” “Month‑end takes 15 days,” “We have three versions of P&L”). From that single sentence, the team expands it into a working system the client owns: documented workflows, reconciliations, controls, and governance. Within ~60 days, a live data model and automated reconciliations establish provable lineage and seal the first cracks; by ~120 days, 60–80% of repetitive workflows are automated, exceptions are routed and tracked, investor reporting accelerates, and risk and finance speak to a single source of truth. The Target Operating Model is then institutionalized: the Current State is benchmarked, the Ideal State is defined (provable lineage, automated recs, consistent reporting, embedded controls), and a Future State path is laid out (agentic workflows, real‑time risk) aligned to mandate and culture. OPCO.AI hands back the capability with clean documentation and durable design, and measures success in hard outcomes: fewer breaks, faster closes, cleaner audits, and basis points saved.

Discovery → Solutions: Diagnose, Design, Deploy

Discovery maps a fund’s operating DNA — beyond “the close is slow” to the root causes: lineage gaps, control breaks, and seams between risk, ops, and tech. From that mandate, the team designs toward the Target Operating Model and executes on two tracks at once: tactical fixes (automate the highest‑pain reconciliations, stabilize data feeds, standardize investor reporting) and strategic rebuild (a governed data spine that unifies risk/finance/ops with near‑real‑time NAV/P&L and designed workflows). Embedded throughout is Risk Diagnostics across three dimensions — efficiency, operational robustness, and regulatory resilience — which prioritize what to fix first, how to contain exceptions, and how to evidence compliance. The outcome is a solutions layer the client owns: harmonized with legacy, auditable by design, and ready for AI agents. The outcome is a Solutions Layer the client owns: harmonized with legacy, auditable by design, and ready for AI agents.

Tech & Design: Cockpits, Not PDFs

Most finance UX feels like 1995. OPCO.AI builds cockpit-grade interfaces (Python/React) where clarity is a control. On the AI side, MCP acts like a USB port for agents across fragmented systems, with lineage preserved. The next frontier: agentic AI with DLT-based proof-of-human/intent, so every action is authenticated, authorized, and auditable.

Case Vignettes

• Intraday Risk & P&L. Unified data model delivers intraday risk and P&L; reporting shifts from backward-looking to a real-time control.
• NAV in Days, Not Weeks. Re-engineered workflows + automated recs shrink month-end close dramatically; auditor challenges decline.
• Intelligence over Legacy. An intelligence layer atop Geneva/VPM/Arcesium harmonizes feeds and enables automation without ripping the core.
• Front-Ends that Think. Credit platform teams interact with intuitive, drill-down dashboards; exceptions drive workflows.
• Shadow → No-Shadow, Safely. Reconciliation controls catch breaks early; costs drop without surrendering control.
• PDFs to Data. AI reads and tags loan agreements and notices; covenants, fees, and headroom flow into structured systems.

OPCO.AI’s Risk Parameters Dashboard

OPCO.AI

The Competitive Landscape

Consultants. The Big Four (Deloitte, PwC, EY, KPMG) dominate operations and risk consulting. Their strengths are clear: global scale, regulatory expertise, and armies of specialists who can handle audits, model validation, and remediation at size. The trade‑offs are equally well known: long timelines, junior‑heavy staffing pyramids, and deliverables that skew toward diagnostics and governance artifacts over shipped systems. Large programs can drift into “transformation theater,” where steering committees multiply but on‑desk change lags. Conflicts (audit vs. advisory) and complex RACI structures can also slow decisions, which matters when firms need working controls in weeks, not quarters.

Fund Administrators. SS&C, Citco, State Street, Northern Trust, Apex and others provide NAV calculation, investor services, and regulatory filings at industrial scale. Their advantages: deep control frameworks, established SLAs, and familiarity with the dominant accounting stacks (e.g., Geneva, VPM). Limitations arise when administrators are asked to redesign front‑to‑back workflows: customization is constrained, change management is ticket‑driven, and integration across risk/ops/finance can be slow. The mandate is to keep accurate books, not to re‑architect operating models. As more managers experiment with “no‑shadow” setups, the quality of reconciliations, exception handling, and data lineage becomes a central dependency — and a potential bottleneck.

Technology Platforms. Arcesium, SEI Novus, Allvue, BlackRock Aladdin, iCapital, and Addepar each offer powerful tools, but generally in siloed domains. Arcesium, spun out of D.E. Shaw, provides portfolio accounting and reconciliation systems. SEI Novus emphasizes analytics and manager intelligence for allocators. Allvue delivers fund accounting and investor reporting software to private capital managers. BlackRock’s Aladdin is the dominant risk and portfolio management platform for asset owners, but was designed primarily with the perspective of institutions like pension funds, not buy-side funds. iCapital has built a strong platform for alternative investment access and distribution, while Addepar focuses on consolidated reporting for family offices and wealth managers. Each of these platforms has become essential in its niche, yet most were not built to integrate seamlessly across the full workflow of private credit, hedge funds, and multi-strategy platforms.

Niche AI/Data Startups. Players like Canoe Intelligence, FundGuard, Hazeltree, or 73 Strings focus on automating narrow slices of fund operations. Canoe specializes in extracting and structuring data from fund documents, FundGuard applies cloud-native tech to accounting and NAV oversight, Hazeltree handles treasury and liquidity management, and 73 Strings provides valuation and portfolio monitoring.Each addresses real pain points, but they operate as point solutions. Their limitation is that they rarely integrate across the full operating stack, leaving firms with improved pockets of efficiency but without an overarching infrastructure layer.

Boutique Consultancies. Smaller shops exist, but they lack global reach and the institutional lineage of larger players. Their impact is often limited by scale.

So what? The market is crowded with capable specialists — consultants for governance, administrators for books, platforms for slices of the workflow, and AI point‑solutions for specific tasks — but few offerings bridge them into a unified, governed operating spine. Managers that can prove real‑time lineage, faster closes, exception containment, and audit‑ready automation are the ones winning allocator trust — and, increasingly, capital.

What could go wrong? Operational rewiring rarely ships on Gantt charts. Integrations sprawl, legacy data fights back, and AI systems raise governance questions — explainability, audit trails, and model drift. There’s vendor‑dependency risk, too: moving reconciliations and covenant parsing into a single layer concentrates failure unless controls are designed for containment. The bet here is that disciplined design, human‑in‑the‑loop workflows, and provable lineage can shrink that blast radius.

The Unified Story

Over the last decade, capital migrated: private credit exploded, hedge funds scaled, family offices multiplied, and even banks offshored in search of efficiency. But the pipes never caught up. The operating burden now looks like a bank’s — without bank-grade controls. Data lineage can’t be proven. Valuation governance lacks audit trails. Investor reporting takes weeks, not hours.

The next stress may not come from asset blowups alone — but from operational failures in reconciliation, surveillance, and governance that magnify them.

OPCO.AI’s wager is clear: if Wall Street is going to withstand its next test, it won’t be thanks to another clever trade. It will be because the infrastructure held — transforming operations into alpha.