# AI - The models are fine. The data feeding them isn't. * * * *By Alain Acha* *The views expressed in this article are my own and do not represent the views of my employer.* * * * Every wealth management relationship starts with building a full picture: the mortgage at another bank, the pension from a previous employer, the business account elsewhere, the loan taken out last year. It all gets written down. The file gets updated. Three months later, the client remortgages. Their salary increases. They open a new account. None of that reaches the adviser unless the client actively tells them. The adviser is working from a static record. The client's financial life has already moved on. That is not a failure of process. It is a failure of infrastructure. And it has a name: **Static Finance**, the condition in which a firm's understanding of its clients is permanently behind their reality. * * * ## The problem most firms won't say out loud Static Finance is not just disconnected systems. In many firms, client data does not exist in a structured, usable form at all: PDFs that no system can read, spreadsheets on local drives, notes buried in inboxes. AI won't rescue a firm still running on inbox archaeology. McKinsey estimates that remediation of poor customer data quality consumes around 20% of a typical financial institution's operating spend.¹ That is money spent maintaining broken data, not building on top of it. That is the foundation on which 94% of financial services firms are now deploying generative AI, according to Databricks², and yet the impact is uneven. Some firms are seeing measurable gains. Many are not. Latent Bridge's analysis of AI adoption in banking found that by late 2025, many stalled initiatives had reached the same conclusion: the models were performing as expected. The data feeding them was not.³ That is a different problem altogether. AI does not fail loudly. It fails quietly, by giving faster wrong answers. * * * ## What Live Finance looks like The alternative is not aspirational. It already exists in pockets of the industry. **Live Finance** is a simple idea with a hard implementation: a firm's understanding of its clients updates continuously, in step with their actual financial lives. Not quarterly. Not at the next review meeting. Continuously. In practice, Live Finance requires continuous external data access, event-driven internal systems, and a unified client data model. The operational difference is immediate and specific. A wealth manager sees a client's full external holdings before the review meeting, not after it. A lender approves a loan against live income and spending data instead of waiting three weeks for PDF statements. A compliance team flags a material change in a client's financial position in real time, rather than discovering it when the client mentions it in passing. A client receives a relevant recommendation because the system noticed a behavioural shift, not because someone manually updated a file. None of these are AI breakthroughs. They are the baseline of what AI can do when the data feeding it is live. The firms operating this way do not look more innovative. They look more responsive. In financial services, responsiveness at scale is a structural advantage that compounds. It is considerably harder to replicate than any model. * * * ## What open banking does about it Open banking does not fix bad internal data. A firm that cannot manage its own data will not be saved by an API. But it solves a different and equally important problem: access. It turns external financial data from something episodic and manual into something live and usable. That is what closes the gap between Static and Live Finance. Combined with internal data discipline, it gives AI usable inputs for the first time. The data was always there. Open banking turns it from static recordsinto live infrastructure. Connecting to external infrastructure does not solve internal data problems. It allows firms to start operating with live data while those problems are being solved. The sequencing changes. The need for internal transformation does not. That is why capital in financial services is moving toward the plumbing, not the presentation layer. * * * ## Who is making the structural bet The signal is not any single deal. It is the convergence, and it has been building for years. Visa acquired Tink in 2022. Mastercard integrated Finicity in 2020. Stripe launched Financial Connections in 2022 to give businesses direct access to consumer bank data. Apple integrated open banking into its Wallet app in the UK in late 2023.⁴ These were not isolated bets. They were a pattern: card networks, payments infrastructure, and Big Tech all converging on the same outcome: a world where financial data moves as easily as payments do. The infrastructure layer is not being built. It is already being operated and extended. These firms are not experimenting with APIs. They are making a structural bet: whoever controls the data connectivity layer will sit upstream of every product built on top of it. That is not an infrastructure play. It is a market position play. The firms most at risk are not the laggards who have not started. They are the ones that have started, piloting AI, running proof of concepts, announcing model deployments, while building on top of the same fragmented Static Finance foundations. They are accelerating on the wrong substrate. The gap between their AI ambition and their data reality is not a roadmap item. It is an architectural constraint. And it widens every quarter they do not address it. * * * ## The regulatory floor is rising In April 2026, the US Consumer Financial Protection Bureau's (CFPB) Personal Financial Data Rights rule came into force for large banks, legally requiring them to share consumer-authorised financial data with third parties.⁶ The EU's instant payments regulation is live. The UK has over 16 million active open banking connections.⁷ The strategic question is no longer whether open banking will emerge. It is whether firms are building for the world it creates, or waiting until the world has already been built around them. * * * ## The honest counterargument Data privacy concerns are real, and they are not merely regulatory friction to be optimised around. Consent infrastructure is still immature in most markets. Many clients remain unaware of what they are authorising when they share financial data, and the gap between what firms say they do with that data and what they actually do is not always small. There is also a legitimate and unresolved question about value distribution. If the firms that primarily capture value from open banking are the aggregating platforms, not the clients generating the data, then the promise delivers considerably less than the headline suggests. History in platform economics gives little reason for optimism here. These are real constraints on the pace of adoption. Compliance, consent, and client transparency need to be built into data strategies from the start, not retrofitted later. But they are constraints on the how and the how fast. They are not constraints on the direction. * * * ## The constraint, answered Most firms in financial services have a well-funded AI strategy and a data estate that has not materially changed in a decade. They are investing in models that operate on Static Finance and expecting Live Finance outcomes. The constraint is the data. Most senior leaders already know this. The harder truth is that knowing it and fixing it are not the same problem. The data infrastructure debt sitting inside most large financial institutions is not an awareness problem. It predates current leadership. It is embedded in vendor contracts, core system dependencies, and architectural decisions made when the business case for Live Finance did not exist yet. A typical transformation programme does not have the timeline, the mandate, or the political surface area to resolve it cleanly. That is why the infrastructure layer is evolving around them. Visa, Mastercard, Stripe, and Apple are not waiting for incumbents to fix their data problems from the inside. They are building and extending the infrastructure that makes those problems solvable from the outside, positioning themselves upstream of every financial product that depends on it, and leaving incumbents increasingly downstream. For wealth management firms, the infrastructure layer is being built regardless. The firms that connect to it early and intentionally will have a structural advantage over those that treat it as a future consideration. The open banking integrations your technology partners have built will determine whether your advisers work from live data or static snapshots. That is a competitive decision about the quality of advice your firm can deliver, not a technology procurement one. The firms moving toward Live Finance are not waiting for a perfect internal transformation. They are connecting to the infrastructure that already exists and building on top of it. The rest are still debating the roadmap. The infrastructure layer is already being owned. * * * ## About the author Alain Acha is a strategy and transformation leader with eight years of experience working across some of the world's largest asset managers and wealth managers in the US and UK. He focuses on the intersection of financial services, technology, and emerging infrastructure. The views expressed in this article are his own and do not represent the views of his employer. * * * ## Sources 1. McKinsey — Next-gen banking success starts with the right data architecture (2025): [https://www.mckinsey.com/capabilities/tech-and-ai/our-insights/tech-forward/next-gen-banking-success-starts-with-the-right-data-architecture](https://www.mckinsey.com/capabilities/tech-and-ai/our-insights/tech-forward/next-gen-banking-success-starts-with-the-right-data-architecture) 2. Databricks — 8 AI and Data Trends Shaping Financial Services in 2026: [https://www.databricks.com/blog/8-ai-and-data-trends-shaping-financial-services-2026](https://www.databricks.com/blog/8-ai-and-data-trends-shaping-financial-services-2026) 3. Latent Bridge — The Most Important AI Trends for Banks in 2026: [https://www.latentbridge.com/insights/the-most-important-ai-trends-for-banks-in-2026-what-will-actually-change-in-operations-compliance-and-risk](https://www.latentbridge.com/insights/the-most-important-ai-trends-for-banks-in-2026-what-will-actually-change-in-operations-compliance-and-risk) 4. Financial IT — Apple Wallet, Open Banking and SoftPOS: The Triple Threat in Payments (2023): [https://financialit.net/blog/openbanking-softpos/apple-wallet-open-banking-and-softpos-triple-threat-payments](https://financialit.net/blog/openbanking-softpos/apple-wallet-open-banking-and-softpos-triple-threat-payments) 5. TechCrunch — Stripe launches Financial Connections for customers to pull banking data (2022): [https://techcrunch.com/2022/05/04/stripe-launches-financial-connections-for-customers-to-pull-banking-data-used-to-power-transactions-automatically/](https://techcrunch.com/2022/05/04/stripe-launches-financial-connections-for-customers-to-pull-banking-data-used-to-power-transactions-automatically/) 6. CFPB — Personal Financial Data Rights Rule, April 2026: [https://www.consumerfinance.gov/rules-policy/final-rules/personal-financial-data-rights/](https://www.consumerfinance.gov/rules-policy/final-rules/personal-financial-data-rights/) 7. Open Banking UK — Statistics 2026: [https://www.openbanking.org.uk/](https://www.openbanking.org.uk/)