Lendi plots AI-native future with MongoDB data revamp

Lendi Group is targeting June 2026 to become "fully AI-native" as it rebuilds its data foundations and expands the use of AI agents across its mortgage and property finance services.

The Australian fintech is moving away from legacy relational databases and consolidating data in a new operational data layer built on MongoDB Atlas. Lendi Group reports $107 billion of home loans under management.

The program follows the merger of Lendi and Aussie Home Loans, which left the combined business with a fragmented technology estate. Over time, it accumulated more than 500 deployable components and ran a mix of relational databases, such as PostgreSQL, alongside non-relational systems.

This structure was costly to operate and difficult to maintain. It also limited the consistency and agility needed for AI services that depend on current, complete customer information.

Data consolidation

An AI-driven mortgage platform depends on combining multiple types of information. Lendi Group says it needs to unify property data-such as valuations, suburb trends, and geospatial information-with finance data, including credit reports and Open Banking feeds. It also needs behavioural signals such as customer goals, prior interactions, and product usage.

Compliance adds another constraint. Operating in a regulated environment, Lendi Group says security and privacy requirements shaped the AI platform's design from the outset, and the platform must adapt as regulations change.

MongoDB's document database model sits at the centre of the new architecture. The platform also includes vector search features used to retrieve information for some AI applications. Lendi Group says this reduces the need for a separate vector database and lowers stack complexity.

The operational data layer project began with a decision to standardise how teams model and exchange data. Lendi Group has adopted a document-first approach and a unified schema strategy, which it says standardises data contracts across domains and makes complex data structures easier to manage.

Engineering choices

Scalability was another requirement. Lendi Group expects data volumes to grow as it expands its use of AI, and points to MongoDB's horizontal sharding as a way to scale without adding operational burden.

Security and compliance controls also sit within the database layer. Lendi Group says it is using built-in security features and a continuous audit trail to support data lineage and accountability, strengthening governance, traceability, and regulatory compliance.

"There simply wasn't another option that offered the flexibility of the document model and the power of MongoDB's integrated, AI-ready data platform," said Will Hargan, Senior AI Systems Engineer at Lendi Group.

Developer impact

Lendi Group says the new operational data layer has accelerated development of AI features, reporting a 40% improvement in time to market compared with the legacy architecture.

It cited Lendi Guardian as an early example, saying the feature moved from development to launch in a 12-week cycle as a test of the new platform's release pace.

Chief Technology Officer Devesh Maheshwari said the shift has also changed how teams build and operate software.

"MongoDB has given us operational simplicity and incredible developer velocity for AI features. The successful launch of Lendi Guardian demonstrates the speed and quality of what we're able to do now," said Maheshwari.

Over the longer term, Lendi Group plans to expand the use of AI agents across routine parts of the mortgage process. It highlighted document checks, follow-ups, and rate monitoring as areas where software agents could take on more work, leaving human brokers more time for complex structuring and customer guidance.

The business described a shift from "human-motion" work to "agentic-motion" work, with software agents handling routine operational steps. It also used the term "elastic workforce" to describe the mix of people and AI agents it expects to develop.

Lendi Group's plan to become AI-native by June 2026 puts the data layer at the centre of the transformation. The next phase will expand the operational data layer and build more AI-driven services across its ecosystem, including property search, buyer advocacy, mortgage broking, conveyancing, and ownership tools.