Setting the file. One moment.
Skill 29 · Knowledge Base For Startups
Subchapter 29.205
references/learn/general/manual-processes-to-mobile-first-ablecredit-is-reinventing-lending-with-aws.mdMarkdown7 KBView on GitHub
AbleCredit (opens in a new tab) is transforming the way banks and financial companies in India evaluate, approve, and monitor credit applications. Working with AWS, it built a mobile-first, multilingual, AI-driven risk assessment platform which streamlines workflows for customers and has enabled the startup to scale across the country.
Traditionally, lending companies have relied on manual processes to conduct credit evaluations. Agents must visit borrowers in person, take photos, hold open-ended conversations, and write assessments from memory. Without structured logs, standardized questions, and methods for capturing consistent data, the process is often slow and resource-intensive, and the results unreliable.
India’s linguistic diversity creates an additional hurdle, with interviews taking place in more than 10 languages and several dialects. “Two similar loan applications could end up with very different assessments depending on who conducted the field visit,” explains Harshad Saykhedkar, Co-founder and Chief AI Officer at AbleCredit. Finally, agents often complete only two to three cases a day, slowing lending decisions for banks.
AbleCredit’s platform is revolutionizing credit evaluation, democratizing access to finance, reducing processing times, and increasing accuracy. Instead of relying on manual approaches which could take up to eight days, lenders using AbleCredit’s solution can produce an assessment in 30 minutes and a full report within an hour. Field agents can also capture photos, record interviews in local languages, and upload case files through a unified mobile workflow. “With our AI-driven platform running on AWS, banks get consistent and faster assessments, and borrowers benefit from quicker approvals,” says Saykhedkar.
The company has also improved throughput by around 200 per cent for its lending partners. Agents who previously completed two to three cases a day can now process up to 10, helping banks expand their lending operations without adding large field teams. In parallel, lenders using AbleCredit have seen EMI-bounce (failed loan repayment) risk decrease by as much as 38 per cent, reflecting more consistent evaluations across languages, agents, and regions.
To scale its offering nationwide, AbleCredit needed a faster, more consistent way to process multilingual audio and varied document inputs, as well as a cost-effective approach for training and refining its AI models. The startup already ran its core infrastructure on AWS and continued using it as the foundation for its platform. “We’re an AWS-first company,” explains Saykhedkar. “From our web app to our databases and training, everything runs on AWS.”
AbleCredit uses Amazon Elastic Container Service (opens in a new tab) to host both its web application and mobile backend in the Asia Pacific (Mumbai) Region (opens in a new tab), serving multiple non-banking financial companies through a single, multi-tenant architecture. Data for each customer is separated in Amazon Simple Storage Service (opens in a new tab) (Amazon S3) buckets, which store training data, model checkpoints, and evaluation logs. AbleCredit also uses Amazon Rekognition (opens in a new tab) to analyze images collected during field visits, supporting automated checks alongside audio transcription and document review.
Field interviews often shift between languages and regional dialects, making it difficult to extract consistent insights or compare cases reliably. AbleCredit’s platform manages this linguistic variation at scale, with AI agents trained to convert multilingual audio into structured, contextual summaries that lenders can use confidently across regions and portfolios.
Before each training cycle, AbleCredit prepares and cleans multilingual audio data stored in Amazon S3 and uses Amazon Elastic Compute Cloud (Amazon EC2) Capacity Blocks (opens in a new tab) to secure GPU capacity in advance. This gives the team reliable access to the compute it needs to run large, multi-day training jobs easier and with fewer delays. The strongest-performing models from each cycle are stored in Amazon S3 and can be re-used as the platform continues to scale.
AbleCredit relies on a number of AWS managed services to help it handle large volumes of data securely and reliably. Application data is stored in Amazon Relational Database Service (opens in a new tab) for instance, while Amazon S3 retains training datasets and model outputs. Field submissions flow through Amazon Simple Queue Service (opens in a new tab), which helps the platform absorb traffic spikes from agents working across regions and languages. Amazon Simple Notification Service (opens in a new tab) then coordinates each step of processing, from transcription through analysis and finally into a standardized report. Together, these services enable AbleCredit to deliver a unified AI workflow for processing audio, documents, and assessments.
What began as a small-scale platform has grown into a nationwide deployment. “When we launched the platform, we started with a few cases a day. Now, we’re processing a few thousand cases a month across India,” Saykhedkar says. Building on this momentum, the company is exploring additional AWS offerings including a listing on AWS Marketplace.
AbleCredit’s journey reflects that of many startups as they evolve from early builds to operating at scale. AWS Activate (opens in a new tab) is designed to support founders throughout this journey, combining technical guidance and expert resources with AWS Credits (opens in a new tab) which help offset the cost of building and running workloads. Since 2013, more than 350,000 startups worldwide have joined AWS Activate, and the program has provided over US $8 billion in credits to support startup growth. Learn more about AWS Activate and how it can help you build, scale, and grow on AWS.