My Advice to Banks on AI: Symmie Swil of Upvest

I spoke with Symmie Swil, UK General Manager at Upvest, Europe's leading investment infrastructure provider. With nearly two decades across banking and fintech, including roles at Investec and Starling, Symmie shares practical advice for bank executives navigating AI strategy and infrastructure modernisation.

Over to you Symmie - my questions are in bold:

Can you give us an introduction to you and an overview of your organisation?

I'm Symmie Swil, UK General Manager at Upvest. Upvest provides the investment infrastructure layer that financial institutions and fintechs plug into if they want to modernise their product offering and launch best-in-class investment experiences.

Before Upvest, I spent close to two decades across banking and fintech, including Investec and Starling. Today, my focus is helping UK financial institutions modernise their investment stack so they can ship new investment products faster, run them with lower operational costs, and make them ready for AI-driven experiences that meet evolving consumer and regulator expectations.

If you were advising a bank CEO today, what would you say is the single biggest mistake they're making with data and AI?

The biggest mistake is trying to "add AI" on top of infrastructure that still runs on batch processing, manual hand-offs, and siloed systems. It's like fitting a Tesla engine into a Morris Minor - it won't go faster, it'll just break down more expensively.

In the UK this is particularly risky given the emphasis the Consumer Duty places on delivering good outcomes and fair value. Here's a tangible example: imagine trying to offer Targeted Support - which the FCA is actively encouraging - when a customer's ISA is in one legacy system, their SIPP in another, their current account in a third, and their statements are stitched together across systems at month-end. You simply can't give personalised, context-aware insights when you don't have a single view of truth. And you certainly can't automate it.

The fix isn't just better AI, it's better investment infrastructure: API-first and modular with real-time data and event-driven processes. Only then does AI have something reliable to work with.

What's one AI or data capability banks should prioritise in the next 12–18 months, and why?

It's really important for banks to prioritise real-time data functionality that connects end user data to the live investment lifecycle - which means three things:

First, a governed data layer that ingests live events (e.g. order lifecycle, positions, cash movements, corporate actions, reporting outputs) as they happen. Second, API-first infrastructure that makes this data immediately usable in customer-facing apps and analytics tools. Third, workflows that go from insight to action without human intervention such as automated recurring investing, rebalancing triggers, goal-based and personalised nudges, and operational automation.

Where do you see banks overestimating AI, and where are they underestimating it?

Banks overestimate what happens when they layer AI onto legacy infrastructure. You can automate a chatbot that tells customers their ISA transfer is delayed, but you can't actually fix the underlying mess of manual processes and fragmented custody records that caused the delay in the first place.

And there's a couple of things they underestimate.

Firstly, the speed at which customer expectations shift once AI becomes the default interface. We've already seen how instant payments and streamlined onboarding - once revolutionary - is now just expected. The same will happen with investment experiences and wealth guidance. Customers will expect increasing levels of personalisation and advice at low or no cost as table stakes.

Secondly, operational efficiency. With event-driven and real-time functionality, AI can streamline customer service, transfers, and reporting with measurable cost reduction. But only if the underlying data is reliable and flows in real time.

What does "good" actually look like when AI and data are working well inside a bank?

"Good" is when the customer can move from question to outcome in one flow, because data and execution are connected. Or even better - they proactively receive the right information at the right time in the right context without even asking.

A customer opens the app and sees a real-time view of their ISA, SIPP, and GIA performance, with tax-efficient recommendations based on their circumstances. They can act immediately: invest, rebalance, set up a regular auto-investment plan, adjust risk - all through automated workflows that execute instantly, not next week after manual approval.

From the bank's perspective, every action is explainable and auditable, with full data lineage built in rather than cobbled together from spreadsheets when the FCA comes knocking.

Operationally, "good" looks like high levels of automation across the lifecycle: order validation, routing, settlement, custody reconciliation, corporate actions, tax handling, and reporting running with minimal manual touch.

What's the hardest AI or data decision bank executives are avoiding right now, and why?

The hardest decision isn't about which AI model to adopt, but committing to the architectural changes required to make AI genuinely effective.

Executives should start by assessing the status quo across product, technology, and operations, and aligning their transformation target and roadmap with their strategic business objectives. Once the target customer experience they want to deliver has been defined, identifying the specific frictions and limitations standing in the way of it is essential. Based on this, they can agree on the architectural changes needed to deliver on this vision and bring in innovation such as AI to their full customer base.

Modernisation to a modular, API-first investment infrastructure is best done in phases to reduce risk. Retain control where meaningful (e.g. customer interactions such as digital frontend and service channels), outsource commodity functionality for processes that do not lead to product differentiation, and leverage scale through partners.

The banks that win will treat AI as the final layer, not the first. You can't build a skyscraper on Victorian foundations. Only when the investment infrastructure is modernised can AI deliver the outcomes that regulators expect and customers demand.

Thank you Symmie! You can connect with Symmie on her LinkedIn Profile and find out more about the company at upvest.co.