Blog — S&P Global Sustainable1 — 07 July, 2026
From hype to impact: AI agents for banks
How agentic AI is changing what's possible across investment banking, lending, and sustainability, and what it takes to deploy it responsibly.
By Harry Tong and Bernhard Schiessl
Highlights
- Agentic AI is moving from pilot to production in the banking sector, taking on multi-step analysis across ESG data, regulatory frameworks and market intelligence.
- This article covers four agents S&P Global Sustainable1 has launched: Transition Finance, Net Zero Engagement, CSA Benchmarking and Data Center Intelligence.
- Larger banks building their own AI agents can still leverage essential intelligence, S&P Global's data in multi-agent systems or the Model Context Protocol, to add trusted climate and sustainability data into agentic workflows.
This content is provided for informational purposes only. It does not constitute investment advice, research, a recommendation, an offer, or a solicitation to buy or sell any securities or financial instruments, nor does it provide advice on the suitability of any investment or transaction. Any examples or references to workflows, outputs, or use cases are illustrative only. Users should exercise their own judgment and seek independent professional advice as appropriate.
Over the past years, the rapid development of AI agents has created new possibilities for optimizing banking workflows. Bankers can go beyond fixed, task-specific workflows to AI agents that execute multi-step analysis across ESG filings, regulatory frameworks, and market data without a human directing each step. There are considerable risks and change management to be overcome before this value can be materialized.
This article focuses on what agentic AI means for banks. We’ll describe four agents we are now offering to clients, relating this suite of solutions to how banks are adopting agentic AI, a full solution including a user interface or the tools to integrate data with their own agentic builds. To learn more about the foundational concepts behind AI agents, including differentiated data, the Model Concept Protocol (MCP) and multi-agent collaboration, see our introductory article.
AI agents at work in banking
Sustainable finance in banking operates where capital, energy and sustainability converge. The workflows are complex; bankers must navigate financial decision-making while considering energy transition strategies and sustainability risks. Time-consuming analysis and fragmented data sources are common.
AI agents can streamline these processes by reducing friction in tasks, such as evaluating decarbonization plans, assessing physical risks and identifying capital raising opportunities. This technology enables bankers to make more informed decisions efficiently, transforming areas previously hindered by manual research and disconnected information.
Before we dive into four of the workflows our agents help solve, let’s address one of the first questions a front-office banker might ask: Where do the figures and analysis in an agent’s output come from, and are they reliable?
The key point to understand is that the AI agent leverages trusted data; it does not generate figures. Every output from our agents is grounded in deterministic data points from S&P Global's underlying datasets. Where a number appears, it can be traced back to the source disclosure and the methodology used to process it. Where reported data does not exist, modelled or estimated figures are flagged explicitly rather than presented as disclosed. Scope 3 and emerging market coverage gaps are surfaced in the output, not hidden underneath a clean-looking number. This level of transparency and traceability is what makes the agent’s output credible.
The agents get a team to a grounded starting point faster but reserve the final steps for a banker to validate the work. The professional in the room, the human in the loop, decides what to do with it.
Debt capital markets banker: Originating green and sustainability-linked finance
Today’s workflow: Preparing for a green bond origination pitch means assessing the issuer's decarbonization, benchmarking it against sector peers and identifying relevant capital raising risks and opportunities, all before sitting down with a CFO. That research typically takes days of analyst time across multiple data sources.
Transition Finance Agent
- Assesses the issuer's decarbonization targets and transition plan against sector peers and Paris Agreement-aligned pathways
- Draws on several S&P Global datasets: ESG scores, environmental data, transition risk, business involvement screens and regulatory intelligence
- Identifies green and sustainability-linked financing opportunities and flags credibility risks in the issuer's decarbonization plan
- Produces pitch-ready content structured for a conversation with the CFO, prospects, clients or investors
The result: Pitch deck content is ready in minutes, creating more time to gather feedback and prepare for the origination meeting.
M&A banker: Advising on data center infrastructure deals
Today’s workflow: Advising on a data center deal means stitching together energy market data, physical risk assessments and demand forecasts from multiple providers under deal timeline pressure. Standard comparative analysis does not capture the nuances of the market, and getting a clear picture can take longer than deals plan for.
Data Center Intelligence Agent
- Assesses power procurement arrangements and grid constraints at the asset level
- Evaluates physical climate risk for each asset in the portfolio
- Maps where AI compute demand growth is likely to concentrate geographically over time
- Draws on S&P Global market data, physical risk datasets and energy infrastructure intelligence in a single pass
The result: Market intelligence structured for an investment decision within minutes. The deal team focuses on judgement and scenarios rather than data assembly.
Equity capital markets banker: Positioning an issuer's ESG story ahead of a transaction
Today’s workflow: Preparing an ESG comp set for an IPO or equity issuance means pulling ESG scores manually, building a peer set, cross-referencing index criteria in a spreadsheet and trying to shape a narrative before the first client call.
CSA Benchmarking Agent
- Benchmarks the issuer against sector peers using data collected in the S&P Global Corporate Sustainability Assessment (CSA)
- Surfaces where the issuer leads, where the gaps are, and what the ESG story looks like to the target investor base
- Flags where no CSA data is available rather than substituting a proxy score
The result: A peer benchmarking output produced before the first client call. Building on the results, the banker can walk in with an initial defensible view, not a work in progress, ready to discuss valuation and demand.
Risk and portfolio teams: Managing financed emissions exposure
Today’s workflow: Assessing Scope 3 financed emissions exposure means pulling data from one source, checking sector decarbonisation pathways from another, mapping counterparties manually and producing something coherent for a credit or portfolio committee. It takes days, and the picture is only as complete as what the analyst managed to find.
Net Zero Investment Agent
- Applies Net Zero Investment Framework (NZIF) methodology to map each counterparty and sector against Paris Agreement-consistent pathways
- Identifies alignment gaps and where transition risk is most material, organized by sector and client
- Distinguishes reported data from modelled estimates
The result: A structured briefing on a company covering more counterparties, with each figure traced back to an actual data point, for the credit or portfolio committee. The analyst reviews and interprets; the agent has done the assembly.
Building AI agents
The larger banks we work with are building their own agentic ecosystems, and large organizations with proprietary datasets and deep domain expertise have good reasons to invest in their own AI infrastructure. For those banks, S&P Global sits at the data layer, providing verified sustainability intelligence, physical risk datasets and market data that their agents consume AI ready data retrieval mechanisms (e.g. MCP, as explained in our previous article).
But building and maintaining a production-grade agent is not a one-time cost. It requires ongoing model maintenance, data quality management at scale, governance infrastructure and specialist talent across AI engineering and financial services compliance. For many institutions, and for many use cases even inside the largest banks, that equation does not add up.
The market for AI tools in financial services is growing fast, and several specialist products are emerging that wrap AI interfaces around sustainability and market data. However, there is a meaningful difference between an AI tool built on top of another company’s data and one built by the organisation that produces the data itself.
When a regulator or a client asks where a number came from, the answer needs to be complete. With S&P Global, the data, the methodology and the agent share the same provenance. The audit trail does not stop at the AI layer and point to a third-party source. It runs all the way back to the data, disclosures, analytical framework and the team behind them.
The value of AI agents is only realized if the governance underneath them is right. Banks operate under obligations spanning data privacy model risk management and fast-moving AI-specific regulation. There are a few things we consider when building AI for banking:
- Client data stays protected. Counterparty information, deal terms and portfolio holdings cannot flow into third-party AI tools without proper data residency, access controls and auditability in place.
- Outputs are auditable. Every agent conclusion must trace back to verifiable data. Regulators expect it and clients demand it. Our architecture grounds every output in deterministic, reviewable data points.
- Humans stay accountable. For important decisions, recommendations and regulatory filings, there must be a professional in the loop. AI accelerates the work, but it does not absorb any liability.
Sustainability teams: What changes with AI agents, and what does not?
Sustainability professionals in banks spend a significant portion of their time on data assembly: pulling figures, cross-referencing frameworks, building comp sets and mapping against regulatory criteria. It is skilled work, but it is not where the expertise adds most value.
Agents take on much of the manual lift:
- Data assembly across emissions, ESG scores, sector pathways, and regulatory screens
- Peer benchmarking against Corporate Sustainability Assessments (CSA) for specific issuers or transactions
- Initial mapping against Net Zero Investment Framework (NZIF) and other frameworks
- Flagging data gaps and distinguishing reported from modelled figures in any output
What remains, and what becomes more important, is interpretation, client advisory and staying ahead of the disclosure or reporting frameworks the agents need to align with. The sustainability team’s role becomes more expert, not smaller. This is what people mean when they talk about a future in which teams learn how to “manage” AI agents.
Where do we go with this next? The agents described here are not prototypes. They represent the kind of grounded, workflow-native AI that is beginning to change how banking teams operate. Individually, these agents save hours and days, but collectively they scale number of deals, client conversations and counterparty screenings without compromising governance.
The institutions that will capture the most value will not be the ones with the most agents. They will be the ones that started with the right workflows and built the governance to match.
Find out more about our HorizonsAgents here.
Disclaimer:
S&P Global and its affiliates disclaim all responsibility and liability for any decisions made or any harm, damage, or losses arising from reliance on or use of any content for any purpose or end-use, including in connection with beta versions and features, all of which are provided "as is" and "as available" and used entirely at your own risk. By using the generated content, you accept these terms. Some or all of the content made available to you pursuant to S1 AI Agents was generated by an AI tool in accordance with our Terms of Use. It may therefore contain errors, omission, inaccuracies, hallucinations, biases, inconsistencies, or outdated information. AI-generated content is provided solely for informational purposes and should not be considered a substitute for human-generated content. Additionally, it is expressly understood that the product, its use, its data, and any information it generates do not constitute, and are not a substitute for, professional consulting advice. AI generated content must be independently reviewed, verified, and approved through standard compliance procedures including compliance with applicable laws. Such content does not constitute investment advice and should not be relied upon or treated as a substitute in any regulatory or decision-making process, including those involving securities, financial products, or related determinations. The AI tool does not create or modify any ESG ratings, scores, or benchmarks, including in connection with any S&P Sustainable1 data or products, and any references to any ratings, scores or benchmarks are for informational display only and must not be construed as advice or endorsement.