Stellar Innovations cuts mortgage review time 40% with CLARAA on AWS

Stellar Innovations provides services to title and mortgage companies across the United States. Mortgage lending and title insurance are complex, data-intensive processes involving multiple types of documents. Although the industry has spent the past 10 to 15 years moving from paper-based processes toward digital workflows, reviewing and validating documents still requires significant manual effort.

A single mortgage package can contain hundreds of documents that teams must classify, reconcile against source data, and validate against individual loan conditions. The challenge becomes more pronounced as mortgage volumes fluctuate with interest rates. When rates fall, application volumes can rise quickly before additional reviewers can be hired, trained, and brought to full productivity. Service providers may also need to respond to increased demand from several clients at the same time, putting pressure on existing capacity and turnaround times.

Stellar Innovations saw an opportunity to reduce manual touchpoints and help teams manage changing volumes more efficiently while keeping experienced reviewers focused on decisions that require judgment. “Faster processing matters not only to lenders and service providers, but ultimately to borrowers trying to reach the closing table. At the same time, we need to protect sensitive borrower information and maintain appropriate governance as AI use grows across the mortgage industry,” says Steve Greenfield, CMB, SVP–mortgage solutions at Stellar Innovations.

To address these challenges, Stellar Innovations built CLARAA, a configurable loan intelligence platform that processes information across the loan cycle, including origination, fulfillment, underwriting, closing, and due diligence. 

“We had been using AWS internally for years, and we valued having the technology we needed 
in one environment, backed by close support from our AWS account and solutions architecture teams,” says Rajesh Kumar Karthigesan, chief solutions architect at Stellar Innovations. The AWS team also worked with Stellar Innovations to review its architecture, identify opportunities to reduce costs, and provide guidance on model selection and optimization. 

CLARAA moves loan packages through a structured workflow for document classification, extraction, reconciliation, validation and audit. Users upload packages through the CLARAA application portal, hosted on Amazon Elastic Compute Cloud (Amazon EC2). The platform classifies documents and reconciles multiple versions before using Amazon Textract and foundation models available through Amazon Bedrock, together with Stellar Innovations’ own algorithms, to extract the fields required for each loan. Confidence scoring helps determine when results require human review, while a configurable rules engine evaluates the file and returns an outcome for each rule, including pass, fail, referral for manual quality control, or a flag for follow-up when required documentation is missing.

Stellar Innovations uses multiple foundation models for different parts of CLARAA rather than relying on a single model. These include Amazon Nova models and open-weight models. The team selects models based on the level of intelligence required for each task and continuously compares their performance, cost, and turnaround time. This lets Stellar Innovations adjust its model choices as new options become available and as existing models evolve. 

CLARAA runs within a private Amazon Virtual Private Cloud (Amazon VPC). Loan documents are stored in Amazon Simple Storage Service (Amazon S3), while AWS Lambda microservices orchestrate the workflow. Stellar Innovations also uses Amazon WorkSpaces virtual desktops for its business process outsourcing and operations teams to review exceptions within the private environment. The platform maintains an audit trail of decisions and keeps human review within the same secured AWS environment, supporting Stellar Innovations’ requirements for governance and traceability. 

“CLARAA is a structured, auditable framework driven by AI. We need documented evidence of what happened and why a decision was made, because these processes can be audited years later,” says Ravindra Nag, chief automation officer at Stellar Innovations. 

With CLARAA on AWS, Stellar Innovations is accelerating the review of complex mortgage files. The business estimates that a simple manual process at scale can be reduced by approximately 40 percent, with opportunities to further shorten turnaround time as the platform continues to improve. CLARAA also reduces manual touchpoints by 90 percent, helping teams spend less time keying and reconciling information across loan files.

A post-closing quality-control file can contain hundreds of pages and thousands of data points. Instead of relying on the typical “stare-and-compare” process associated with mortgage quality control, where an auditor manually compares every piece of information, analysts can focus on exceptions surfaced by CLARAA while keeping human reviewers in the loop where further analysis is needed. When exceptions require review, Stellar Innovations’ teams have a 3–4-hour SLA for corrections.

“When an underwriter is looking at a mortgage file that can be hundreds of pages long, CLARAA lets them focus on exception management instead of manually comparing hundreds or thousands of data points. The human stays in the loop, but they can be more productive,” says Greenfield. By recognizing and cross-referencing data points, CLARAA highlights exceptions and inconsistencies across the loan file, helping strengthen risk management.

CLARAA is already being used in various forms by several large lending and real estate companies, including Fortune 500 companies. Customers use the platform for processes ranging from checking loan packages to extracting data and feeding it into loan origination systems. Stellar Innovations says CLARAA has helped customers reduce turnaround times and is also helping shift the company from primarily providing business process outsourcing services to working with customers as a technology partner.

“On AWS, we can continue evolving CLARAA as customer volumes and technology requirements change. We continually evaluate foundation models to optimize cost and turnaround time, while working with AWS as new models and capabilities become available,” says Nag.