Financial services spent the last two years running artificial intelligence as a pilot: a chatbot bolted onto a support queue, a proof of concept in a single trading desk. This week two very different announcements showed the pilot phase ending. Google Cloud shipped a packaged agentic AI platform built specifically for banks and capital markets firms, and a five year old private markets data company raised a $20 million round to expand the AI tooling it already sells to 150 investment firms managing $400 billion. Neither is a chatbot experiment. Both are infrastructure, sold as infrastructure, to institutions that plan to run on it.
The shift: from pilot to platform
The tell is not that banks are using AI. They have been for years, in fraud models and credit scoring. The tell is that the vendors serving them have stopped selling “AI features” and started selling AI as a managed layer that sits underneath existing workflows, with its own connectors, its own audit trail, and its own governance controls built for regulators rather than for consumer trust and safety teams.
That distinction matters because a feature can be switched off. A layer that a bank’s research desk or a private equity firm’s portfolio team has rebuilt its process around cannot be, at least not without real switching cost. Financial institutions are being sold permanence, not a demo.
Advertisement
Google Cloud builds the financial research layer
Google Cloud on Tuesday unveiled Gemini Enterprise for Financial Services, a purpose built agentic AI package for capital markets and corporate banking. The centerpiece is a Google managed Financial Research agent, paired with more than 50 specialized skills, 13 data connectors into providers including FactSet, S&P Global, LSEG, Moody’s and SEC Edgar, and a growing roster of third party agents from partners such as S&P Global and Dun & Bradstreet.
The launch is not theoretical. CME Group and Deutsche Bank are already using it, with Deutsche Bank named as a design partner for the Financial Research agent specifically. Google Cloud said the broader Gemini Enterprise platform is already live inside BNY, Citi Wealth, Lloyds Banking Group, Macquarie Bank and insurer Signal Iduna.
“Financial professionals are looking for an AI platform that doesn’t lock them into any one model or ecosystem, connects to the IT systems they use every day, and is highly secure and compliant,” said Thomas Kurian, CEO, Google Cloud.
The design choices back that framing up. Every output from the Financial Research agent carries confidence scores, a stated methodology, data snapshots and source citations, an explicit answer to the objection that has kept large financial institutions from putting general purpose AI in front of anything client facing: no traceable lineage for the answer it gives.
Private markets get the same treatment
The same pattern showed up a rung down the market. Standard Metrics announced a $20 million Series B led by 8VC, with Salesforce Ventures and Spark Capital among the participants. The company has run investor relations infrastructure for venture and private equity firms since 2020, and says it has grown roughly 20x since its Series A, with more than 10,000 portfolio companies and 150 firms managing over $400 billion in assets now on its platform. Thirty percent of the current Forbes Midas List are customers.
The new capital is not funding a pivot. It is funding an expansion of AI document parsing, an on platform AI analyst, and interoperability with the Model Context Protocol, the same open standard Google Cloud’s connectors lean on. Two infrastructure vendors serving very different parts of finance are converging on the same architecture: models that plug into a firm’s existing data rather than requiring the firm to feed a chatbot.
“Our mission at Standard Metrics is to accelerate innovation in the private markets,” said John Melas-Kyriazi, co-founder and CEO of Standard Metrics. “This Series B round is a vote of confidence from investors who believe, as we do, that the firms who embrace AI now will be the ones defining the next era of the private markets.”
What it means for the finance leader
For a bank, insurer or investment firm evaluating these platforms, the practical questions have shifted. It is no longer “does this AI tool work,” it is “what does it plug into, who audits its output, and what happens to our workflow if we later want to switch providers.” AI agents have already moved from chat interfaces into the middle of finance this year across banking, payments and trading. What Google Cloud and Standard Metrics are doing is formalizing that move into a purchasable, auditable product category, rather than leaving each institution to stitch its own agent stack together.
That formalization has a cost side too. A managed research agent with 13 licensed data connectors, or a portfolio platform with MCP interoperability, is a recurring commitment, not a trial license. Finance leaders should expect procurement conversations that look more like a core banking system purchase than a SaaS add on, with the same scrutiny of data residency, model lock in and exit terms.
How to evaluate these tools
Three questions separate durable AI infrastructure from a feature that will not survive a budget review: does every output carry a traceable source and methodology, does the platform connect to data the institution already licenses rather than requiring new feeds, and does switching vendors later mean rebuilding a workflow from scratch or simply repointing a connector. Not every vendor entering this market will answer all three well. The two that announced this week did, which is precisely why they are worth watching as the shape of what comes next. Institutions still running AI as a pilot program should treat that as a signal to set a decision date, not to wait for a better one. Some European banks have already made the opposite bet, building sovereign models in house rather than buying a managed layer; the split between build and buy is becoming the real strategic question in financial AI, not whether to adopt it at all.
Source: Google Cloud
