Three ways to put AI into action: insights from our ‘Practical AI Workflows’ webinar

By Jilbert El-Zmetr and Jack O’Connor

The rise of AI is transforming how investment firms operate. Highly manual and repetitive tasks can now be automated, freeing teams to focus on higher-value portfolio decisions and ultimately drive alpha. While our previous blog – following our ‘Practical AI Workflows’ webinar – focused on finding the right tools, conducting proper assessments, and implementing frameworks that prioritize data management, compliance, and security, this article explores how those tools can be put into action. 

Here, we highlight three practical use cases where AI can help investment firms increase efficiency, and generate greater capacity across their operations.

Trade Breaks and Reconcillation

A familiar challenge for firms in the fund management space is the misalignment between what a broker executed and what the order management system (OMS) recorded for trades from the previous day – whether it’s price, quantity, or another detail. Identifying these trade breaks and performing reconciliation processes is very time consuming, arduous and heavily manual, severely limiting the ability to focus on other alpha-generating portfolio decisions.

Fortunately, AI tools can help automate this process. For example, daily trade reports can be fed into tools such as Power Automate to collate OMS reports and prime broker emails. These files can then be processed with ChatGPT using a predefined prompt to identify breaks, gaps, or issues, summarize the findings, and automatically email the relevant stakeholders in the front and back office.

These workflows can be enhanced further with a third-party technology provider to structure data and unlock deeper insights. This could include:

• Adding a spreadsheet-based audit stored in SharePoint, listing the issues identified by ChatGPT, and providing a historical record for compliance and audits?
• Developing a template file to standardize subject lines for counterparties
• Incorporating attachments or links in summary emails that reference the original reports or communications

The biggest advantage of this approach is that it often leverages tools already available within a firm’s existing toolkit, while saving hundreds of hours annually that would otherwise be spent manually reviewing reports and emails.

Quarterly Performance Reviews and Communications

Drafting quarterly performance reviews and investor communications can take a significant amount of time – not only compiling all the data, but also writing the report in the tone investors are accustomed to.

AI agents, such as Copilot, can be used to simplify and accelerate the processes by prompting them to draft an investor newsletter automatically. To work effectively, a couple of key inputs are needed: investor data should be stored centrally (e.g., in SharePoint or Outlook), and historical emails should be stored in an Outlook folder.

This setup allows Copilot to easily compile the information and learn the tone and structure of past communications, enabling it to generate curated, personalized messages for different investor types, such as allocators or pension funds.

Investment Research

Investment research in itself takes time. But the process is made even more manual when data needs to move between the front office and research teams. AI can significantly streamline these workflows, but this requires data to be organized and connected to the tool from the outset. 

For example, earnings call summaries and transcripts should be stored in easily accessible locations – such as Teams or SharePoint. Email-based news should also be stored in dedicated Outlook folders, while Excel models should reside in SharePoint for consistency across platforms. 

Structuring email-based research received from brokers, market data providers, or platforms such as/including FactSet, can be more challenging. Increasingly, firms are using connectors to agentic AI tools to integrate their Office 365 environment with external data platforms, enabling natural language queries across their subscribed datasets.

Additionally, accessing historical data for research is slightly harder to integrate for AI workflows, as most firms use either an all-in-one cloud-based order and portfolio management system or separate OMS and PMS platforms. This data typically lives in a database, often SQL, which does not integrate directly with Copilot within Office 365. To address this, periodic CSV or data file extracts can be pulled into SharePoint, enabling aggregation with other research data.

Once this data is organized,  firms will be easily able to use an AI agent (e.g. Copilot) with targeted prompts to surface the insights investment managers are looking for. This includes:

• Forecasting performance using historical returns
• Obtaining rapid and relevant summaries of news on specific stocks
• Back-testing historical decisions by adjusting variables such as/including interest rates or volatility

For more information on how Siepe can help you implement AI tools and workflows across your operations to streamline processes and drive efficiencies, contact [EMAIL].