Building Trusted AI Systems in Financial Services From Strategy to Scale - with Zar Toolan of Edward Jones
🎯 Summary
Podcast Episode Summary: Building Trusted AI Systems in Financial Services From Strategy to Scale - with Zar Toolan of Edward Jones
This 27-minute episode features Zarr Toolan, General Partner and Head of Data & AI at Edward Jones, discussing the critical steps financial services firms must take to move AI initiatives from stalled pilots to scaled, trusted production systems. The core narrative emphasizes the necessity of strategic alignment, robust governance, and human-centered change management.
1. Focus Area
The discussion centers on AI Transformation in Financial Services, specifically addressing barriers to production adoption, establishing responsible AI governance, navigating the build vs. buy vs. partner decision, and managing the human element (change management and talent strategy) required for scaling AI.
2. Key Technical Insights
- Data Sanctity as the Coin of the Realm: The success of any AI strategy hinges on the underlying data governance and the “sanctity of the data.” Clean, sovereign data that can be transposed across multiple systems is the foundational competitive advantage.
- Shift from Use Cases to Scalable Platforms: Organizations should stop measuring progress by the sheer number of AI use cases (Generative or Agentic) and instead focus on building and measuring scalable, repeatable platforms and patterns that generate measurable strategic outcomes.
- Multi-Modal Delivery: AI solutions must be designed for multi-modal delivery—serving end consumers, clients, and back-office operations in structured and intentional ways, governed by strict data and model oversight.
3. Business/Investment Angle
- Strategic Alignment is Non-Negotiable: AI transformation cannot exist as a separate initiative; it must be directly and intentionally integrated into the overall corporate strategy and vision.
- Build vs. Buy Guideline: For capabilities that do not drive significant competitive advantage, buy the solution. Reserve building or strategic partnership for areas that constitute the firm’s “secret sauce.”
- Partner Selection Criteria: When choosing partners, look beyond technical systems integration expertise (vertical/industry knowledge). Crucially, assess their horizontal expertise in human-side change management to ensure adoption across the workforce.
4. Notable Companies/People
- Zarr Toolan (Edward Jones): The featured expert, providing practical insights from a major individual investor-focused financial services firm.
- Emerge AI Research: The host organization, which conducts executive surveys and hosts thought leaders like the CIO of Goldman Sachs.
5. Future Implications
The industry is shifting into the “Intelligence Age” or “Agent Age,” which is predicted to be a faster and more massive transformation than the shift from the Information Age to the Internet Age. Success will depend on organizations’ ability to leverage data to create continuous feedback loops that improve processes and experiences, rather than just optimizing current states.
6. Target Audience
AI/Tech/General Business Professionals in highly regulated industries, particularly Financial Services Executives, CIOs, CDOs, and AI Strategy Leaders responsible for scaling AI adoption and managing governance risks.
Comprehensive Summary
Zarr Toolan of Edward Jones provided a strategic roadmap for embedding AI into the core of financial services operations, moving beyond pilot purgatory. He identified two primary reasons AI initiatives stall: lack of direct alignment with corporate strategy and the failure to establish clear guiding principles for utilization.
Toolan outlined five essential guiding principles for responsible AI: Human-Centeredness, Accountability, Transparency, Trustworthiness, and Inclusivity. These principles are crucial for navigating regulatory scrutiny and building user confidence, especially as agentic systems become more prevalent.
The discussion then pivoted to the Build vs. Buy vs. Partner dilemma. Toolan stressed that this decision must be rooted in data and model governance. He advocated for a pragmatic approach: buy commodity functions, and build or partner strategically for proprietary competitive advantages. He used the analogy of the “Peloton,” urging firms to stay in the lead pack by leveraging established platforms for the 70% of needs, while focusing internal resources on the differentiating 30%.
Perhaps the most significant theme was the human element of AI adoption. Toolan argued that firms over-index on the tool set (technology) and under-index on mindsets and skill sets. Successful adoption requires making AI an exceptional, seamless experience in the flow of work—ideally, users shouldn’t even realize they are using AI on the front end. He emphasized the need to focus change management efforts on the “movable middle” (70-80% of the workforce) to drive mass adoption.
Finally, Toolan addressed talent strategy, advocating for a portfolio approach (full-time, contingent, offshore) based on role duration and required skill sets for the next 2-3 years. A critical component of this strategy is knowledge transfer—systematically downloading tribal knowledge from long-tenured employees into scalable training and AI systems to bridge onboarding gaps for new talent, regardless of their employment status. Ultimately, the conversation underscored that data quality remains the single biggest barrier across all industries, and mastering data architecture is the key to unlocking sustainable AI outcomes.
🏢 Companies Mentioned
đź’¬ Key Insights
"organizations should buy solutions that maintain parity with the market, but build or partner where they can create true competitive advantage."
"aligning AI initiatives directly with corporate strategy is non-negotiable."
"Knowledge transfer is one of the most critical pieces that we see in this. You have, especially in the, not just in the knowledge management knowledge worker space, this is true of industrials and manufacturing. This is true across the board of you have a long-tenured associate, 20 to 30 years plus that have a lot of knowledge in their head about how things work. They know the shortcuts, they know where all the systems are, where they work, where they don't, they know the workarounds. How are you downloading that knowledge into the skills mapping?"
"The one thing that is consistent across those back to where you started, Matthew, on data, companies have got to get their data right. That is the biggest barrier that we see from we talk about legacy tech debt a lot, like the actual systems and processes and back office as behind the scenes regardless of industry. But it's really the underlying data."
"What we really need to focus on are the mindsets and skill sets, because that's where change efforts are failing in the space, is over-indexing on the tool set and the underestimating, under-indexing on what's going to take from a mindset and skill set standpoint."
"It really needs to start with a shift from use cases to scalable patterns and platforms."