Three Key Takeaways From Siepe’s Practical AI Workflows Webinar

AI workflows are increasingly being adopted by investment managers, rapidly transforming how the industry approaches its operating models, and driving efficiency at every level. Yet AI is still in its infancy, and firms continue to navigate questions around which tools to use, how to implement them effectively, and how to ensure they deliver real business value. For alternative investment managers, successfully addressing these challenges can create a meaningful competitive advantage and keep them in step with the evolving ecosystem.

In our latest “Practical AI Workflows” webinar, we explored how fund managers can adopt and implement AI technologies effectively, the key considerations to keep in mind, and the practical use cases for leveraging them in a way that automates and enhances day-to-day workflows. Here are our key takeaways:

Navigating the AI tool landscape effectively

There are an abundance of AI tools available to investment managers. Each is suited for different use cases and levels of complexity. Our webinar touched on the following tools:

• AI Assistants (e.g. Copilot, Gemini): Best for enhancing daily workflows – drafting emails, summarizing documents, generating reports, and quickly retrieving information, especially when integrated with your firm’s security.
• General Purpose Chat (e.g. ChatGPT): Widely known for research, multi-format content generation (text, image, audio), and broader analytical tasks. While not as tightly embedded as AI assistants like Copilot, they can be integrated via services such as Power Automate or Foundry. Business and Enterprise versions enable deeper data connections, and provide control over model training and secure use of company data.
• Enterprise Development Platforms (e.g. Azure AI Foundry, AWS Bedrock): Cloud-based environments built for firms operating at higher complexity. Users typically need deeper expertise in machine learning and AI. These platforms provide access to multiple foundational models and enable integration with APIs, internal and external databases, and existing technology infrastructure.
• Custom LLMs: Using custom LLMs is where firms fine tune foundational models with their own proprietary data sets. This is typically used for highly specific, data-dependent tasks, such as fund-specific risk modeling and pattern recognition on historical trade performance.
  

By leveraging these AI tools effectively, investment managers can streamline daily tasks, enhance analysis, and make more informed decisions.

Successful implementations start with the business problem and selecting the right tools

Only around 5% of AI pilot programs achieve rapid revenue acceleration. Of the 95% that fall short, most failures stem not from the AI model itself, but from a fundamental learning gap within the organization. This means it’s critical for firms to first define a clear business problem, before simply deploying “AI for AI’s sake”. Without a proper assessment, businesses risk falling short and delivering very little return on their investment. AI might not be needed at all – you might only need an improved automated workflow.

As mentioned above, tool selection is equally important. Solutions need to align with a firm’s technical and operational complexity for achieving desired outcomes, and addressing specific needs. For investment managers, this includes looking at how data integrates and flows across systems, understanding upstream and downstream dependencies, and ensuring compliance requirements are embedded into the workflow.

Data management, compliance, and security all need to be prioritized

These are three non-negotiable elements when determining a firm’s AI strategy. Data management and integrity are integral to eliminating duplicate data sources and ensuring quality, completeness, and consistency across systems. This requires firms to assign clear data ownership of workflows – for example, defining who owns investor data – so AI solutions can automate processes and deliver reliable, real-time data feeds.

As with many workflows in the Investment management industry, compliance protocols must also be followed to ensure products adhere to covenants and regulatory standards. Audit trails and explainability should be built into AI-driven research and trading decisions. At the same time, human oversight remains essential, particularly when reviewing automated investor communications before distribution. Additionally, clear internal policies governing the use of AI must also be established and enforced to ensure staff adhere to appropriate and compliant practices.

Security underpins both data management and compliance. Data moving between systems needs to be encrypted both in transit and at rest to prevent unauthorized access. Access controls can be restricted to different data sets based on least privilege permissions. Regular risk assessments and penetration testing should also be conducted to ensure that security controls remain robust and aligned with evolving standards.

—

At Siepe, we understand that AI presents a significant opportunity to help investment managers scale, streamline workflows, and make more informed decisions. But this requires a strategic approach that clearly identifies the business problem; selects the right tools that meet your operational complexity; and embeds robust data management, compliance, and security practices. By doing so, firms can leverage these technologies to enhance their technology stack and stay ahead in a rapidly evolving industry.

We’d like to thank everyone who was able to attend our webinar – we really appreciate your questions and insights. To learn more about how Siepe partners with investment managers to navigate and leverage AI solutions across a wide range of strategies and technical complexity requirements, contact us at help@siepe.com.