How Redding Ridge Uses AI to Assemble Credit Portfolio Data

Michael Manapat introduces Rowspace, which works with financial firms to reconcile fragmented data using agents. He is joined by Jack McDonald, COO of Apollo-affiliated structured-credit manager Redding Ridge.

McDonald says Redding Ridge manages structured credit, primarily collateralized loan obligations or CLOs. He describes the firm as a separate operating company and says it managed about $50 billion at the time of the talk.

He describes thousands of incoming data points across many funds and investments. Active CLO management depends on having current, consistent information available when portfolio managers make decisions.

Much of the information is private and fragmented across documents and systems. McDonald says Rowspace works alongside the team to understand calculations, workflows and which source should be authoritative.

His first example is portfolio construction. CLOs have specific tests and payment waterfalls embedded in indentures and legal agreements, which portfolio managers must take into account.

He says a manager previously spent roughly two to three days building a mock portfolio of 200 to 400 credit names in Excel. In his team's experience, current information and the new workflow can produce an initial portfolio in minutes.

McDonald characterizes that output as roughly 80–90% complete, after which portfolio managers apply judgment and refine it. These figures describe his team's experience, not an independently verified guarantee of accuracy or speed.

The workflow brings together document libraries, trade and compliance systems, and pricing data. The point is not just generating text, but assembling information used in a portfolio decision.

His second example is warehouse purchase activity. Previously, staff reconciled purchases and sales across bank statements and sent a weekly recap to management.

McDonald says the reporting process is now repeatable in about a minute in their implementation, with staff performing daily quality checks. He compares it with a former manual process that occupied two or three people for hours each week.

Manapat asks why a general chatbot is insufficient when the output supports trading and portfolio decisions. The practical issue is not just model capability but access to the firm's own validated data and business rules.

McDonald says general-purpose models are useful for research and learning, while their implementation centers on the firm's own data. General models can also be connected to private data; the distinction he emphasizes is how this particular workflow is grounded and validated.

He says one data item may appear in many sources, with only one appropriate for a particular purpose. Rowspace staff sit with portfolio and finance teams to understand the calculations and select or reconcile sources.

The team validates information from trade management, performance systems and external pricing sources before relying on it. Automation reduces manual assembly, but the speaker does not claim that verification is unnecessary.

Asked about Apollo's adoption culture, McDonald describes a willingness to adopt new technology, improve team capability and replace repetitive processes. These are his observations about the organization.

He considers AI adoption in large financial organizations to be early. Banks and asset managers have complicated middle- and back-office systems, and applications must account for that complexity.

At Redding Ridge, he says the immediate focus is making routine work more efficient so staff have capacity for other projects. He also expects some research and junior-support tasks to change over time.

Manapat asks whether AI will become a common productivity tool or a lasting competitive advantage. McDonald thinks it can improve competitiveness when combined with a strong process and reliable information.

In active and private markets, he sees value in knowing available liquidity, prices and other relevant facts promptly. Inaccurate information, by contrast, can flow through decisions to the organization and its investors.

Looking ahead, McDonald wants confidence in information shared with investors without spending excessive time on manual checks. This is a goal for the team, not a claim that errors have been eliminated.

He also wants technology to become part of the team's everyday thinking. Time saved on routine tasks can be used to ask why a process exists, whether it should change and what tools would help.

His advice to builders is to recognize that financial organizations are highly nuanced. A workflow that appears unusual may exist for a reason; spend time with the people who actually perform the work.

McDonald attributes progress with Rowspace to the team's sustained presence with the portfolio and finance staff. Manapat closes by emphasizing the human and technical work required to make such systems useful.