CRED, India’s most rewarding fintech platform for the creditworthy, has built CRED codelens, a living intelligence layer that gives every line of code the full context of CRED’s platform, architecture, services, and guardrails. With CRED codelens, the company has created a single source of truth for every function, and an agentic layer that orchestrates 400+ specialized agents to accelerate decision-making and shipping velocity. Amazon Bedrock routes tasks to the right model and enables enterprise-grade controls. CRED codelens is a flagship effort of CRED’s AI labs initiative.
With 20+ business verticals, 2,000+ repositories, and 500+ microservices, engineering complexity at CRED compounded faster than documentation or process could keep up. CRED codelens makes systems self-explanatory and distributes context gathered from Slack, Confluence, Jira, and other tools, as well as the code itself.
Starting from the customer-facing app, it traverses microservices to produce end-to-end request call graphs that map how different applications, services, APIs, and user journeys connect and communicate across the platform. Every code commit automatically triggers analysis, generates service and flow-level documentation, and indexes it into a vector database within 30 minutes.
On top of this foundation, an agentic layer orchestrates specialised agents for codebase Q&A, PR review, code generation, infrastructure debugging, and test creation. Today, CRED codelens works alongside team members on every line of code in the organisation, reducing what would take weeks to build, test, troubleshoot, and deploy into hours.
Amazon Bedrock gives CRED a single governed API across frontier models with automatic routing to the best-fit model per task, and enterprise-grade compliance and cost controls built in. The platform runs on Amazon Elastic Kubernetes Service, with supporting infrastructure for vector storage, observability and a registry of secure MCP servers spanning engineering, data, and product functions.
Within 12 months of launch, CRED codelens has become CRED’s connective tissue. Data and analytics teams run cost optimisation agents. Business and operations teams use it for campaign advisory, customer support, and talent workflows. Security and compliance teams conduct threat modelling and compliance audits through the same platform.
With 500+ active users and 400+ agents running across engineering, data, product, and operations teams, the productivity impact is significant: engineers ship new features and resolve bugs 4x faster, CI/CD pipeline debugging shows 40% effort savings. CRED codelens evaluation shows 74% code segment coverage, 92% endpoint coverage, and 65% helpful answer rates.
Kiran Jagannath, Head of FSI and Conglomerates, AWS India and South Asia, said, “CRED has reimagined how engineering teams operate at scale by building CRED codelens on Amazon Bedrock — turning their entire codebase into living knowledge. What stands out is how CRED has moved beyond isolated AI use cases to create an enterprise-wide intelligence layer, with 400+ specialized agents accelerating decisions across engineering, product, and operations. This is a powerful example of how India’s most innovative companies are using AWS’s generative AI capabilities to fundamentally transform how software is built and shipped.”
Swamy Seetharaman, AI enabler at CRED, said, “When every team member has instant, accurate knowledge of how the system works, decisions improve before a single line of code is written. CRED codelens transforms our codebase, conversations, and context into living knowledge that teams can query, reason over, and act on, compounding the value of every engineering decision. Amazon Bedrock gave us the flexibility to build this at scale across models, while maintaining strong controls around security, governance, cost, and measurable ROI.”
