Newgen Software

NewgenONE began as a content and process automation platform, and its lending applications run on low code workflows, document management and business rules that predate the AI layer. Models now carry specific jobs. The credit decisioning engine runs rule batteries and model batteries in parallel and weighs more than 200 parameters per application.

Marvin automates routine tasks and answers questions about documents, LumYn analyzes customer behavior to find revenue opportunities, and the early warning system watches portfolios for delinquency patterns. Agent Studio and a data science studio let banks build and deploy their own models and agents. Without the models, NewgenONE remains a working loan origination and collections system.

Complex credit applications are flagged for human review and arrive with pre analyzed risk, AML checks and suggested counter offers, and the decisioning engine supports agent review and sign off. Simpler ones can go straight through, and Newgen says the engine automates up to 90 percent of underwriting steps. Agentic Shield sets operating boundaries and limits for agents, applies business policies and engagement rules, and monitors agent performance.

Every decision is logged with its timestamp, rule version, model output and any human intervention. Newgen does not give the threshold that sends a case to a person, or say whether a decline can be issued without review.

Each credit decision comes with a breakdown of contributing factors, such as credit score, income stability and spending patterns, and with risk probabilities such as probability of default. The logs keep the rule version and model output behind every decision. For models a bank builds itself, the data science studio offers performance monitoring, champion and challenger models, automated model selection and explainable AI.

No accuracy figure accompanies Newgen's own models, and there is no mapping to a model risk standard. Its claim that approval rates often rise by 2 to 5 percent without added credit risk comes without a method.

Newgen's investor presentation for the quarter to June 2026, filed with the Indian stock exchanges, reports about 500 active customers in 79 countries, 101 customers billing over ₹50 million and banking at 63 percent of revenue. It also lists a $1.71 million retail loan origination deployment in the Philippines, without naming the bank.

Lending brochures cite results such as abandonment down 60 percent, disbursement up 45 percent and straight through processing up 50 percent, with no customer, method or period attached. A Forrester Total Economic Impact study on its site reports a 371 percent return on investment.

Agents work on the organization's own data, with no external data used for training, and act only on consented, customer approved data. The platform detects and redacts personal data automatically, filters outputs for hallucinations and off brand content, and manages consent, with data residency commitments. These commitments come from Newgen's product material rather than a contract or data policy, and no retention period for prompts, outputs or training data is given.

The privacy policy, revised in January 2026, covers Newgen's websites and its applications, services and tools, and names the contracting Newgen entity as data controller. It keeps data as long as operations require, allows transfers to affiliates and third parties, including in India, without naming a transfer mechanism, and names the GDPR as its only law. Personal data, it says, is not used for automated decision making. Borrower data that banks process through NewgenONE, where Newgen works for the bank, falls outside the policy, and there is no subprocessor list.

Newgen's ISO 27001:2022 certificate, issued by Intertek Certification under UKAS accreditation, certificate number 0199925, is valid until 17 December 2027 and can be downloaded from its company page. The scope covers the design, development, delivery and support of software products, solutions and cloud offerings across eight sites in Noida, Delhi, Chennai, Navi Mumbai and Gurugram, and the certification dates back to December 2009.

Newgen also holds SOC 1 Type 2 and SOC 2 Type 2 attestations covering all five trust services principles, it says, and its managed cloud is certified to ISO 27017 and ISO 27018 and compliant with the PCI Data Security Standard. The SOC reports themselves are not posted.

As a software supplier, Newgen holds no lending license and does not set out a regulatory position for its products. Its banking material discusses what regulators expect, transparency, fairness and auditability, and it writes explainers on the Reserve Bank of India's scale based regulation for non bank lenders. The trade finance agent supports UCP 600 checks on documentary credits. No supervisor, register entry or statute is tied to the lending suite itself.

Newgen says its application risk model reduces manual intervention and bias and supports fair, optimized lending rates, and that diverse datasets are used to minimize bias in its agents. Agentic Shield and the Trusted AI material set out governance without a fairness method. There is no subgroup testing, disparate impact analysis or list of the attributes its credit models weigh, on an engine that issues approve, deny and counter offer decisions. Declined applicants get the factor breakdown each decision carries, and no adverse action process beyond it is described.

Decision logs keep the timestamp, rule version, model output and any human intervention behind every credit decision, and each decision carries its contributing factors, which gives a lender what it needs to reconstruct and challenge an outcome. No product terms, warranty or indemnity covering the AI are public, so who bears the loss when a decision or an extraction is wrong is not known. Newgen does not describe a route for an affected borrower to contest a decision.

A reasoning hub built with LangGraph sits under the agentic AI, with a retrieval engine on a vector database and domain specific fine tuned large and small language models. No model provider, model family or hosting arrangement is named for them, and there is no public subprocessor list. Marvin and LumYn are presented as Newgen's own generative AI models, without saying what they are built on. Corporate lending reads financial statements with Evalueserve's AI based OCR engine, under a partnership announced in December 2023.

The credit decisioning engine connects to credit bureaus, KYC and identity verification, fraud and AML services, consented open banking data, core banking, CRM and data warehouses, mostly without naming them. One core banking system is named. Newgen is a certified Mambu partner, its lending applications pair with Mambu for underwriting, disbursement and portfolio management, and Mambu lists the relationship in its own partner directory.

Corporate loan origination takes financial statements through Evalueserve's OCR engine into a spreading template. A developer documentation portal is public, and decisions reach online, branch and mobile channels through APIs.

NewgenONE can be deployed in the cloud, on premises or in a hybrid setup, so an institution can meet its own regulatory requirements. The AI can run on premises or in a private cloud, with data residency and sovereignty commitments. Newgen also runs a managed cloud, which it says is certified to ISO 27017 and ISO 27018 and compliant with the PCI Data Security Standard. Specific hosting regions are not listed.

Deals are priced by quote, for NewgenONE, its lending applications and its agents alike. Investor material reports deal values in aggregate, such as a $1.71 million retail loan origination deployment in the Philippines, which shows the size of an engagement but gives a buyer no rate to price against. Whether licensing runs per user, per application or per module is not public.

Banking produced 63 percent of Newgen's revenue in the quarter to June 2026, and its lending material is split by segment: retail lending, housing finance, commercial and SME lending, captive finance and leasing, collections and trade finance. The captive finance solution, launched in December 2024, covers lending, leasing, inventory financing and invoice discounting for captives and non bank lenders. Regional sites serve Singapore, Australia, the Middle East and Africa. In Southeast Asia, ASEAN Business Partners resells it across Indonesia, Malaysia, the Philippines, Singapore and Thailand, and insurers and government agencies run on the same platform.