For one of the largest U.S. wealth management firms, customer trust had always been the foundation. But with over 36 million active accounts and an ambitious growth trajectory, even a strong foundation can begin to strain.
Behind the scenes, more than 15,000 service agents were supporting a growing client base – armed with over 2,000 internal documents stored in a sprawling SharePoint repository. Agents were expected to locate, interpret, and respond to real-time client queries using this system.
It wasn’t broken. But it was slow, inconsistent, and costly to scale. Query resolution times averaged 4 – 5 minutes, and accuracy varied depending on who was at the keyboard. Every additional account added pressure to expand the service team – a model that couldn’t stretch much further.
So, the firm decided: stop scaling people. Start scaling intelligence.
Rethinking the Model: From Search to Response
Persistent Systems was brought in – not just to improve the search process, but to eliminate it. The solution wasn’t to make agents faster at navigating documents. It was to remove that need entirely.
Together, they built a GenAI-powered assistant capable of understanding natural language, retrieving the most relevant information, and returning answers that were not only instant but compliant and contextual.
The engine behind it? A tightly orchestrated pipeline that combined Google Cloud’s Vertex AI for document indexing, OpenAI’s LLMs for summarization, and a feedback loop that continuously improved output based on agent ratings. New documents synced automatically. No one needed to maintain the system by hand.
It was deployed across the firm with minimal disruption and maximal ambition.
The Outcome: What Used to Take Five Minutes Now Takes Seven Seconds
Within weeks, the assistant scaled from an initial cohort of 500 agents to over 15,000. Adoption didn’t require mandates. It spread because it worked.
Agents went from manually searching PDFs to simply asking questions like they would in conversation. And the numbers told the story:
- Query resolution time dropped from 4–5 minutes to just 4–7 seconds
- Answer accuracy rose above 93%, based on enterprise-verified documentation
- The system handled over 16,000 interactions daily, across all tiers of agents
- Service delivery became uniform, no more variations based on agent experience
- The firm could grow toward 50M accounts without expanding headcount
For the agents, this wasn’t just a tool, it was an equalizer. Junior staff could now handle complex queries with confidence. For the business, every query became a moment to learn, improve, and deliver better.
But Was It Just a Speed Fix? Not Even Close.
Yes, the assistant was faster. But what it truly changed was structure. The GenAI layer didn’t just make support more efficient, it made it future-ready.
- Compliance was de-risked, thanks to consistent, audit-ready answers
- Senior agents were freed from routine escalations, redeployed to strategic roles
- It laid the groundwork for automation in other areas: onboarding, advisor enablement, and knowledge operations.
It also positioned the firm well ahead of its peers. According to McKinsey, applying GenAI to customer care can deliver 30%–45% productivity gains in the next few years.
So, What Made It Work?
Plenty of enterprises talk about GenAI. Few get it right at scale. What made this engagement stand apart was execution: Persistent didn’t drop in a model and walk away. They owned the full stack – from UI and data pipelines to AI orchestration and cloud deployment. Built entirely on Google Cloud, the solution scaled seamlessly with zero downtime.
Most importantly, it didn’t require heavy internal lift. The client contributed a product owner. Persistent handled the rest.
What’s Next: Expanding the Intelligence Layer
The success of the assistant has catalyzed broader transformation across the enterprise. New initiatives are already underway:
- Real-time, voice-enabled support for contact centers
- Domain-specific LLMs fine-tuned on proprietary financial data
- Automated compliance alerts embedded into responses
- GenAI-powered coaching for frontline agents
Each capability builds on the same foundation – creating network effects from what was once just a search function.
Conclusion: Not Just Better Support but Smarter Business
This was never just about reducing query time. It was about reshaping how knowledge powers customer interactions. In doing so, the firm turned its support function into a strategic capability – one that scales with confidence, learns in real time, and performs with precision.
With Persistent’s GenAI fabric in place, they didn’t just automate answers. They rewrote how answers are delivered.
From minutes to moments. From legacy search to living intelligence. This is GenAI at enterprise scale.
