Table of Contents
How St. James’s Place Transformed Data Operations with Maia, the AI Data Automation Platform
In financial services, trust is everything. Decisions are only as good as the data behind them, and speed can’t come at the expense of governance. For St. James’s Place (SJP), one of the UK’s leading wealth management businesses, the challenge wasn’t a lack of data – it was how quickly and confidently that data could be turned into insight and action.
SJP regularly sends customer surveys and gets thousands of responses, each one rich with feedback that could inform better service delivery and client retention. But manual analysis meant the team were only able to process a small fraction before insights went stale – leaving competitive intelligence and the opportunity to improve client experience on the table. Meanwhile, critical ETL pipeline migrations were consuming scarce engineering capacity, slowing platform modernization.
St. James’s Place needed to prove AI could accelerate data work while maintaining strict security and governance standards. In a proof of concept, they tested Maia, the AI Data Automation platform, on two critical challenges: sentiment analysis of client surveys (consuming 4,000 hours annually) and ETL migrations that were bottlenecking their platform consolidation.
Delivering 1,300% Faster Analysis and Modernizing Data Pipelines
Early results exceeded expectations. Maia reduced end-to-end sentiment analysis to around 16 hours – a 1,300% efficiency gain – and cut ETL migration effort by roughly two-thirds. The POC demonstrated that governed, in-house agentic AI could deliver significant productivity gains, creating momentum for broader adoption and freeing engineering capacity for SJP’s SAP and AI roadmap.
Speed and Trust Drive Every Client Decision at SJP
SJP’s mission centers on delivering trusted, one-to-one financial advice that helps clients plan confidently for their future. Achieving that at scale depends on data that is accurate, timely, and accessible across the business.
To meet rising demand for faster insights, stakeholders needed actionable intelligence sooner. Engineers were being asked to do more with the same resources. And any AI initiatives had to meet rigorous standards for security and trust.
“We all know the power of AI and data,” says Kelly Maggs, Divisional Director for Data Architecture Platform and Engineering. “But we need to roll it out in a secure, well-governed way. Trust is key – speed can’t come at the expense of control.”
The Bottleneck: Manual Work Slowing Strategic Progress
SJP faced two parallel challenges limiting their ability to move at pace:
- Insight delivery: Thousands of client survey responses arrived regularly– qualitative feedback that could provide the opportunity to improve client experience and lifetime customer value. Manual categorization and sentiment analysis meant weeks of lag time, inconsistent interpretation, and the inability to spot trends until they were already old news.
- Platform evolution: Like many enterprises, SJP operated multiple ETL tools across their environment. Consolidating pipelines was essential for long-term efficiency and maintainability, but manual rewrites consumed days per pipeline. Engineers spent time translating legacy logic instead of building the future platform.
Together, these created a familiar tension: the business needed faster insights and modernization, but traditional approaches couldn’t keep up.
Testing Maia: The AI Data Automation platform
With manual workloads creating bottlenecks, SJP designed a proof of concept to test whether agentic AI could deliver real productivity gains within their security and governance requirements.
They selected Maia, the AI Data Automation platform, for evaluation. Maia works directly within the data platform, understands context, and executes complex tasks end to end. It can build, explain, iterate, and document pipelines, all within the organization’s existing, secure environment.
“We wanted to look into how we could use Maia to potentially improve the productivity of our engineers – enabling them to focus on more strategic, in-depth work while Maia focuses on the actual development of pipelines,” Kelly adds.
POC #1: Turning Voice of Customer into Actionable Insight
Client feedback is one of SJP’s most valuable data sources. Surveys regularly capture free-text responses ranging from a single word to detailed commentary on client experiences. Historically, analyzing that data relied on manual categorization, making it difficult to apply consistent sentiment scoring or track trends over time.
SJP tasked Maia with building a secure, in-house sentiment analysis pipeline within their cloud data warehouse. Maia orchestrated the full workflow while keeping all data inside SJP’s governed infrastructure.
The pipeline:
- Categorized responses into primary, secondary, and tertiary themes
- Applied consistent sentiment analysis across each category
- Leveraged large language models programmatically for scale and accuracy
- Produced structured outputs ready for downstream analytics and reporting
The result: What historically required around 4,000 hours of manual effort was completed in 16 hours end to end – a 1,300% operational efficiency gain. The speed of processing demonstrated opens the door to more regular surveys and a deeper understanding of evolving client needs. These structured outputs highlight the possibility of tracking sentiment trends over time and provide a framework that could enable teams to incorporate voice-of-customer insights into future strategic decisions.
POC #2: Accelerating ETL Migration and Platform Consolidation
The second test focused on a different kind of bottleneck. Kelly’s team was exploring consolidation of multiple ETL tools into a single platform to improve efficiency and reduce long-term complexity. Traditionally, this kind of migration requires engineers to manually translate existing jobs, validate logic, and rebuild transformations – often taking days per pipeline.
Kelly says, “We recognized that platform consolidation would help us to reinvest in the team, to enable us to build out more on the SAP and AI roadmap tomorrow.”
With Maia, existing ETL files were ingested and converted into individual transformation logic far more quickly. Maia handled much of the repetitive, error-prone work, allowing engineers to focus on validation and higher-value design decisions.
The result: ETL migration effort was reduced by roughly two-thirds, turning days of work into hours. This not only proved the potential to accelerate consolidation but also demonstrated how engineering capacity could be freed and reinvested elsewhere.
What the POC Revealed: AI Productivity Gains Can Be Real
Maia’s impact was clear;
- 1,300% efficiency gain in sentiment analysis operations
- Two-thirds reduction in ETL migration effort
- Faster insight delivery to business stakeholders
- Freed engineering capacity without additional headcount
- Validation that agentic AI can work within strict governance requirements
Maia demonstrated that the large productivity gains often associated with AI aren’t hype – when applied thoughtfully and governed properly, they’re achievable.
After seeing Maia in action, Kelly’s initial skepticism gave way to excitement. “The big productivity numbers you hear about AI can actually be real,” Kelly says.
What’s Next: Building Momentum for Broader Adoption
For SJP, the POC with Maia demonstrated that agentic AI can augment human expertise rather than replace it. What began as two targeted proof points has created momentum for a broader model of trusted data delivery.
The team is now planning to run controlled comparisons pairing Maia with human engineers to build confidence that Maia Team can reliably handle pipeline development. This measured approach lays the groundwork for broader adoption, enabling SJP to scale trusted, governed AI across its data platform and reinvest engineering capacity into strategic initiatives like SAP modernization and AI-driven insights.
Maia isn’t just speeding up data operations – it’s creating a blueprint for how enterprises can operationalize AI safely, effectively, and at scale.
