About the Company
Credora is a financial technology company that develops alternative credit scoring solutions using
data such as mobile-money activity, transaction behaviour, and airtime usage. Its approach is
designed to provide additional financial behaviour information that can support credit assessment
processes.
Customer Challenge
As Credora’s credit assessment activities grew, analysts needed to gather information from multiple
sources when preparing applications for review. This could include application information,
transaction behaviour, mobile-money activity, airtime activity, existing credit scores, financial
indicators, and applicable credit policies.
Before working with Scyda Technologies, analysts could spend significant time collecting and
organising this information before beginning their assessment. The preparation process also required
analysts to identify missing information, review inconsistencies, perform calculations, and locate
relevant policies, creating repetitive work that could slow the overall assessment workflow.
Partner’s Solution
Scyda Technologies implemented an internal agentic AI Credit Assessment Preparation Agent using
Amazon Bedrock AgentCore, powered by Anthropic’s Claude Sonnet 4.5 model. The solution assists
authorised credit analysts by coordinating approved data-retrieval and calculation tools to prepare
applications for human assessment.
Amazon Bedrock AgentCore Runtime executes the assessment agent, while AgentCore Gateway
provides controlled access to approved business tools. AgentCore Identity manages the agent’s
identity and authorised access to business resources, while AgentCore Memory can maintain
relevant application context during multi-step assessment workflows.
AWS Lambda provides controlled functions for retrieving permitted application and financial
behaviour information and performing approved deterministic calculations. Amazon DynamoDB
supports the solution’s application and assessment data requirements, while Amazon S3 and
Amazon Bedrock Knowledge Bases provide approved credit policies, assessment procedures, and
supporting documentation.
Amazon Cognito and Amazon API Gateway provide the analyst-facing application and authentication
layer. AWS IAM, AWS KMS, Amazon CloudWatch, and AWS CloudTrail provide security, access
control, monitoring, and audit capabilities.
The agent produces an assessment preparation package containing relevant application information,
financial behaviour, calculated indicators, missing or inconsistent information, applicable policy
considerations, and supporting evidence. The system does not autonomously approve or decline
credit applications. Final lending decisions remain with authorised credit analysts.
Results and Benefits
● Credit analysts can initiate application preparation through a single internal interface.
● The agent reduces repetitive manual work involved in gathering and organising assessment
information.
● Approved financial data and policy information can be brought together into a structured
assessment package.
● Deterministic calculations are performed through controlled application functions rather than
being delegated to the language model.
● Analysts receive clearer visibility into missing or inconsistent information requiring attention.
● Human review and final credit decision-making remain within the established credit
assessment process.
About the Partner
Scyda Technologies is an AWS Advanced Partner providing cloud and AI solutions that help
organisations modernise business processes, develop intelligent applications, and build secure and
scalable workloads on AWS.