The rapid integration of Artificial Intelligence (AI) into various sectors, including financial planning, presents a dual-edged sword, offering unprecedented efficiencies alongside significant potential risks. While headlines often grapple with existential threats posed by advanced AI, everyday users encounter more immediate concerns such as AI "hallucinations" – erroneous or fabricated information. Brad Joudrie, CEO of Conquest Planning, a prominent financial planning software firm, is proactively addressing both ends of this risk spectrum within his company’s platform, particularly through its integrated AI agent, SAM (Strategic Advice Manager). Joudrie’s approach emphasizes a rigorous methodology designed to mitigate inherent AI risks like hallucination, agentic misalignment, and data sovereignty, thereby empowering financial advisors with AI-driven efficiencies without compromising client safety and trust.
The pervasive nature of AI technologies necessitates a diligent and informed approach from industry leaders. Joudrie asserts that a fundamental aspect of safeguarding against AI’s potential pitfalls lies in comprehensive due diligence and the implementation of robust control mechanisms. For Conquest Planning, this translates into building AI agents that are inherently auditable and traceable. "One of the ways that we as an industry can protect ourselves is by understanding and doing the right amount of diligence," Joudrie stated. "In the work we’re doing that’s connected to agents, one of the things that those agents can stand on is that the auditability and the traceability that exists in the calculations that we’re building." This commitment to transparency is central to Conquest Planning’s strategy for deploying AI responsibly in the sensitive domain of financial advice.
Controlling for Hallucination and Error in AI
The specter of AI hallucination, where AI models generate incorrect or nonsensical outputs, poses a direct threat to the integrity of financial plans. These seemingly minor errors can cascade into significant flaws, jeopardizing clients’ long-term financial well-being. Joudrie highlights that SAM’s design prioritizes auditable and traceable calculations precisely to combat this issue. This feature allows financial advisors to meticulously verify the AI agent’s work, cross-referencing its outputs with the underlying data inputs and the algorithms used.
"AI agents have become a popular means of expediting workflows and introducing new angles of discussion between advisors and clients," Joudrie observed. "While Joudrie sees why that popularity is growing, he stresses the need for the underlying data that the agents use to be verifiable by the advisor. Without that clarity the ‘small’ error of a hallucination can become a core fault in the plan and pose real long-term risks to the client."
Financial plans are not static documents; they are dynamic tools that must evolve with clients’ changing circumstances. The emphasis on traceability within Conquest Planning’s platform ensures that plans can be revised and updated with confidence. By enabling advisors to trace SAM’s contributions back to the initial data and calculations, any modifications can be made without introducing core errors that could have originated from an AI hallucination. Joudrie confidently asserts that this meticulous process means Conquest’s engine "doesn’t hallucinate."
A significant contributor to hallucination in broader AI models is their reliance on vast, unfiltered datasets scraped from the internet. To circumvent this, Conquest Planning employs a strategic approach by utilizing a "small language model" (SLM). Unlike their large language model (LLM) counterparts, SLMs operate within a confined, proprietary environment. "One of the tools Conquest has elected to use to avoid that is what they call a ‘small language model’ which only pulls from conquest’s native environment," Joudrie explained. This restricted scope ensures that SAM’s outputs are tethered exclusively to the validated data and logic already integrated into Conquest Planning’s system, thereby significantly reducing the likelihood of generating unfounded information. This controlled environment is crucial for maintaining the accuracy and reliability required in financial planning.
Managing Existential and Agentic Risks in AI
Beyond the immediate concern of factual errors, Joudrie and Conquest Planning also acknowledge and actively manage the more profound, albeit less frequent, risks associated with AI, including the potential for agentic misalignment. This refers to the risk that AI agents might act in their own perceived best interests rather than those of their users. This concern has been vocalized by prominent figures in the AI community, such as Dario Amodei, CEO of Anthropic, who has raised alarms about the potential for an "AI-driven apocalypse."
While regulatory frameworks and industry-wide AI policies are expected to play a critical role in addressing these existential threats, Joudrie emphasizes that Conquest Planning has proactively integrated risk mitigation strategies into its design. The company’s commitment to traceability serves as a crucial safeguard against agentic misalignment. "If it appears that the AI agent is making suggestions or plans that serve its interests over the clients, the roots of those suggestions can be identified and corrected through those more traceable pathways," Joudrie stated. This allows for the investigation and rectification of any AI-generated recommendations that deviate from the client’s best interests.
Furthermore, robust cybersecurity principles are foundational to Conquest Planning’s AI deployment strategy. The company recently achieved ISO 42001 certification, the world’s first international standard for an AI management system. This certification underscores their commitment to establishing and maintaining a systematic approach to managing AI risks and opportunities. Complementing this, Conquest Planning has maintained a steadfast data sovereignty policy since its inception. All client data is securely stored within the geographic boundaries of the customer’s location. For instance, Canadian data resides in Canada, UK data in the UK, and US data in the US, all managed through Amazon Web Services (AWS) in each respective country. This localized data storage not only adheres to stringent data protection regulations but also builds trust by ensuring that sensitive financial information remains under the purview of national jurisdictions and is not subject to cross-border data transfer risks.
Addressing the Allure of Insecure AI Tools
A significant challenge Conquest Planning faces is the growing tendency for individual investors to seek financial advice from readily accessible, yet less secure, public AI tools like ChatGPT and Claude. The ease of use of these platforms often overshadows the inherent security risks and the lack of specialized financial expertise. Individuals inputting sensitive personal and financial data into these general-purpose AI models without the robust security protocols and regulatory oversight present in dedicated financial planning software are exposing themselves to considerable danger.
Joudrie recognizes this trend and emphasizes the critical need for user-friendly yet secure platforms. "People are going to at times gravitate to easy, and easy might be plugging some data into a service like Claude or OpenAI to get a financial plan, but that’s really not the best way to approach it," Joudrie commented. He added, "So as a financial professional, you’re not sacrificing any of the speed that these engines can solve for."
To counter this, Conquest Planning aims to provide advisors with a solution that combines the efficiency of AI with unparalleled security and accuracy. Joudrie encourages financial advisors to educate their clients about the risks associated with using public LLMs for financial advice and to guide them towards more secure and reliable alternatives. "With Conquest, I would encourage those financial advisors that are leveraging our service to get your customer or client to stop [using public LLMs for advice]," Joudrie urged.
Moreover, Conquest Planning offers a seamless transition for advisors and clients who may have already utilized public AI tools. The platform allows for the direct upload of data generated from these less secure engines into Conquest’s own LLM. This enables the automatic construction and refinement of financial plans within the Conquest ecosystem, identifying any potential gaps or errors that might have originated from the initial AI analysis. "So we’ve essentially looked to solve for this challenge as well," Joudrie concluded, underscoring the company’s commitment to addressing the full spectrum of AI-related challenges in financial planning. This integrated approach ensures that advisors can leverage AI’s power while maintaining the highest standards of client data security, plan accuracy, and regulatory compliance.
The broader implication of Conquest Planning’s strategy is the establishment of a benchmark for responsible AI integration within the financial services industry. By prioritizing auditability, traceability, and data sovereignty, alongside a commitment to user-friendliness, Joudrie and his team are demonstrating that the benefits of AI can be harnessed without succumbing to its inherent risks. As AI continues its inexorable march into every facet of modern life, the diligence and foresight exhibited by companies like Conquest Planning will be crucial in building a future where technology serves humanity safely and effectively. The ongoing evolution of AI necessitates continuous vigilance and adaptation, and Conquest Planning’s proactive stance sets a precedent for how financial technology firms can navigate this complex landscape.
