Lentra

The platform is origination, servicing and rules software first, and the models sit inside it. Strip them out and GoNoGo, BREx, MultiBureau, 1LMS and the co lending and embedded finance modules still run a lender's applications on configured policy.

The AI reaches the decision path in three places: BREx hosts and triggers machine learning models in real time during underwriting, Lentra MLOps deploys and versions a lender's own risk, fraud and marketing models, and GoGetr applies AI powered extraction, liveness detection and face matching to identity documents. Lentra describes the result as hybrid decisioning that combines deterministic rules with predictive model output.

Most applications can run without a person. GoNoGo claims straight through processing above 90 percent and the consumer durable journey 96 percent, with decisions in under a minute. The control sits in BREx, where lenders configure portfolio level rules that fire in real time: a dealer's applications can move to a hold queue once a set threshold is crossed, and a credit manager can be alerted when approvals within one range of bureau scores pass a set share of the day's total. Which applications reach a person by default, and how automated approvals are sampled afterward, is left to each lender's configuration.

Model tooling is aimed at the models a lender runs. Accuracy, validation or error figures for Lentra's own extraction and face matching models are not published, but the operating controls are. Lentra MLOps keeps a registry of every model, version and environment with rollback, real time monitoring, approval workflows and audit logs, and plugs models into BREx through configurable interfaces.

BREx adds a simulator that runs several models or scorecards offline and compares their results, and lets a lender feed repayment and bureau data back in to check scorecard performance and correct it.

Outcome figures are specific but carry no lender's name. GoNoGo claims a 400 percent decrease in turnaround time, an 18 percent increase in approval rate, a 27 percent cost reduction and straight through processing above 90 percent. BREx claims 23 percent more applications disbursed and a 50 millisecond response time, and the consumer durable journey claims 96 percent straight through processing and five loans a second.

None of these is attributed to a named lender, and no customer story pairs an institution with a result. HDFC Bank has invested in Lentra, which shows a large Indian lender took a stake in the company rather than what the platform delivered for it.

Whether one lender's data can inform models that serve another, and whether Lentra's own models learn from client outcomes, is not stated. The question is concrete here. BREx is built to take repayment information and bureau data back into the engine so scorecard performance can be checked and corrected, and MultiBureau studies the data it pulls to find patterns, such as a bureau returning scores for only part of a lender's clients in one city. What Lentra does describe is the lender's control over its own models, which are deployed, versioned and rolled back through MLOps under role based access control.

Lentra's trust center lists ISO 27701 for privacy information management beside its security certifications and says it carries subprocessor disclosures. The privacy policy on its own site covers website visitors only and is governed by Indian law. A data processing agreement, retention schedule or deletion commitment for the bureau records, know your customer documents and bank statements the platform handles for lenders is not set out publicly. Data is protected with encryption, and the consumer durable journey describes protection for data both in flight and at rest.

Lentra's trust center lists a SOC 2 Type II report and a SOC 3 report beside ISO 27001, 27017, 27018, 27701 and 22301 certifications and a Cloud Security Alliance registry listing, with 12 documents released on request. It also states annual third party audits and penetration testing, a disaster recovery plan, cyber insurance and single sign on for employee access. The auditor and report periods are not named. For a platform holding bureau records and know your customer documents for Indian banks, the reports sit one access request away.

Regulatory requirements are built into the journeys rather than tied to a named supervisor. GoNoGo connects to credit information companies and regulatory bodies, agricultural lending is aligned with priority sector lending mandates, and identity checks run on Aadhaar based electronic know your customer.

Lentra holds no lending license and needs none as a software supplier, and it names no regulator, statute or rule that governs the platform, although the Reserve Bank of India's digital lending rules set obligations for technology providers that sit between a lender and a borrower. The company operates as Lentra AI Private Limited from Pune.

BREx and GoNoGo automate credit decisions with machine learning models running beside lender rules, and GoGetr adds face matching and liveness detection to identity checks. Lentra publishes no testing of either for disparate impact or demographic performance, and no method for producing adverse action reasons. What BREx does offer is segment monitoring inside the day: a lender can track approval rates within a range of bureau scores and alert a credit manager when they drift past a limit it sets.

Lentra publishes nothing on who bears the loss when an automated decision is wrong, and describes no way for a lender or a borrower to challenge an output. A borrower declined in a straight through journey has no route through Lentra to see the factors or correct a record, and the obligation rests with the lender. The platform approves loans end to end in seconds on bureau, bank statement and identity data, and MLOps keeps audit logs of model deployments and approvals.

Lentra names the plumbing more than the suppliers. MultiBureau is built on Kafka, Spring, TensorFlow, Node.js and React, and Karza, Finfort and CreditVidya are named among the third party services it connects. The credit bureaus behind it are described as all the major ones without being named, and no model provider or hosting provider appears, although the trust center mentions subprocessor disclosures. Lenders' own models plug in through MLOps and BREx, so for those the lender is the supplier.

Integration runs through GoGetr, which onboards new third party interfaces through low code configuration, normalizes their payloads into one structure, tracks each call's turnaround and error rate, and falls back to an alternative when a provider fails, at more than 5 million calls a day. 1LMS exposes its functions to customer apps and core systems over web interfaces, though no core banking vendor is named.

Journeys draw on more than 250 integrations, MultiBureau fills every major bureau's requirements from a single entry in the origination system, and named providers include Karza, Finfort and CreditVidya alongside Aadhaar based identity checks.

Hosting is described as cloud native and container based, on a microservices architecture that is always on with dedicated data centers. Where those data centers sit, which provider runs them and whether a lender can choose a region or run the platform privately are not stated. Horizontal scaling and the ability to switch critical vendors such as bureaus or fraud services at runtime are built in for availability.

Pricing follows a software as a service model charged per user or per transaction, with no capital outlay for the lender. No rate, tier, minimum or trial is published, so the billing unit is known while the size of a bill is not. Implementation is pitched as fast: journeys come ready to use, and the loan management system offers single day plug and play setup for its open platform.

Coverage is wide across lending products and concentrated in one country. Product journeys cover personal, home, vehicle, education, professional and business loans, loans against property, overdrafts, equipment and supply chain finance, consumer durables and credit cards, plus agricultural lending that handles Kisan Credit Card reviews and renewals and dairy, poultry and fisheries finance aligned with priority sector lending goals.

Co lending and embedded finance modules extend origination to partner lenders and outside lead sources. The market served is India, with Aadhaar based identity checks and connections to credit information companies built into the journeys.