The Shift to the Intelligent Core: Normalised Data Pools, Natural Language Underwriting, and Human-in-the-Loop Agent Orchestration

At the Money20/20 Europe event, the evolution of core banking infrastructure moved past basic cloud migration to focus on real-time data intelligence. Adrian Congiu, VP of Product Management at Mambu, detailed how the company anchors its technology platform within the data lake architecture, conversational analytics, and autonomous operations layers of the global banking ecosystem. As financial institutions seek sustainable competitive advantages, transforming core ledgers into predictive engines has emerged as a top priority for global banking leaders.

1. Defining the Intelligent Core: Moving from Static Ledgers to Real-Time Predictive Services

The transition from a cloud-native infrastructure to an “Intelligent Core” represents a fundamental change in how financial institutions utilize their primary data assets. Traditional core banking systems function largely as static, backward-looking repositories that simply record historical financial transactions. This legacy structure limits a bank’s agility, preventing it from reacting quickly to live market opportunities or shifting consumer behaviors.

An Intelligent Core changes this dynamic by transforming the central ledger into a highly predictive, personalized engine. This framework allows banks to leverage deep customer data pools in real time, reacting instantly to events as they occur on a client’s account. By turning passive transactional data into immediate operational insights, banks can deliver hyper-personalized financial services exactly in the moments that matter most to their clients.

2. The Mambu Data Lake: Eliminating Integration Bottlenecks to Fuel the Agentic Era

A primary obstacle stalling enterprise artificial intelligence strategies is the extensive data pipeline engineering required to extract, clean, and structure legacy information. Accessing clean, normalized data is an absolute prerequisite for training and deploying reliable AI models. Without it, banks spend months building custom integrations and complex data pipelines before their AI tools can ingest a single file.

The Mambu Data Lake solves this infrastructure bottleneck by serving as a turnkey data extraction platform for core transactions, balances, and account data.

• Pre-Structured Gold Tier Data: The platform bypasses traditional pipeline construction by offering ready-to-consume, normalized data assets directly within its premium “gold tier” data lake layer.

• Simplified Ecosystem Feeding: Banks can immediately extract core financial records and feed them directly into custom AI models or broader internal data ecosystems.

• Multi-Channel Enrichment: This normalized core layer allows technology teams to easily enrich transaction histories with internal channel information and secondary customer data sources, creating a powerful data foundation for autonomous agents.

3. Conversational Analytics: Natural Language Queries and Automated Portfolio Insights

The way banking professionals interact with enterprise data is shifting away from traditional query-based code and rigid analytics dashboards toward a natural language interface. This evolution lowers the barrier to data access, allowing internal banking teams to interact with complex databases through simple conversation.

Mambu utilizes this conversational approach to let risk officers and bank users query the data lake directly using natural language. Instead of writing complex SQL queries or waiting for specialized data analysts, team members can ask direct questions about individual customer files or request broader portfolio performance metrics. This capability eliminates the need for a separate analytics software layer on top of the bank infrastructure. Furthermore, the platform automates these insights, generating periodic risk reports and trend warnings naturally from the underlying data.

4. Back-Office Automation: Product Configuration and Human-in-the-Loop Operations

The operational shift toward “less UI, more AI” is fundamentally changing the daily workflows of back-office administrators and operations teams. Rather than navigating multiple windows and clicking through complex menus, employees can use chat interfaces to complete daily tasks.

Mambu implements this conversational automation across two main operational areas:

Automated Product Configurations

Platform administrators can interact directly with a chat interface to design and launch new financial products. An automated agent assists the user by analyzing the request, determining the optimal configuration path within the Mambu platform, and executing the technical setup automatically.

Streamlining Core Operations

Because most back-office processes eventually interact with the core ledger, Mambu provides a Model Context Protocol (MCP) server that grants autonomous agents secure API access to core functions. This integration reduces the time operations staff spend clicking through disparate software screens to locate data. To maintain absolute security, these workflows are built with comprehensive tracking, observability, and audit logs—ensuring a human remains in the loop to review and approve all automated actions.

Key Highlights from Adrian Congiu:

• The Predictive Shift: Moving to an Intelligent Core allows banks to replace backward-looking legers with predictive, real-time personalization.

• Turnkey Data Extraction: The Mambu Data Lake provides pre-normalized, gold-tier transaction data, saving banks from months of custom pipeline engineering.

• Human-in-the-Loop Governance: Back-office automation uses secure API tokens and strict auditing logs, keeping a human supervisor in the loop for trust.

The post The Shift to the Intelligent Core: Normalised Data Pools, Natural Language Underwriting, and Human-in-the-Loop Agent Orchestration appeared first on FF News | Fintech Finance.