Decisioning platform

How does Banca AideXa, a fintech bank company, use AI Decisioning Platform to support the development of small and medium enterprises?

In an era of banking that continues to transform, innovation is the key to supporting the growth and development of small and medium enterprises (SMEs). What Banca AideXa is doing is an example of this innovation success story.

As they conveyed in the webinar entitled “SME Financing Revolution: How Banca AideXa uses AI Decision-Making Platforms and Open Banking to Support SME Growth”, AideXa is proof of how the technology changes the way financial institutions interact with and empower SMEs. Banca AideXa, a fintech bank focusing on SMEs, has been at the forefront of this revolution in Italy.

With a commitment to simplifying SMEs’ access to credit through a fully digital approach, Banca AideXa leverages artificial intelligence (AI) and the open banking concept, combined with the utilization of a credit decisioning platform. This will result in the creation of a smooth, efficient, and data-driven credit application experience.

This article will recap the insights shared by Banca AideXa regarding how they help SMEs overcome their hurdles, create solutions that suit the needs of SME customers, use data and AI models in their strategy to answer SMEs’ problems, and collaborate with ACTICO as a credit decision platform provider.

Banca AideXa (referred to as Aidexa hereafter) is the first Italian fintech company to focus on the SME sector. Founded in 2020, AideXa was born with a very clear mission, which is to simplify the life of Italian entrepreneurs.

This is why their core business is to lend money with 100% digital experience to small and medium enterprises, leveraging on the opportunities granted by open banking.

This is why their core business is to lend money with 100% digital experience to small and medium enterprises, leveraging on the opportunities granted by open banking.

Aidexa said that the key to their success lies in the presence of data and AI models.

Before going into technical details, we need to know what are the bad and good of SMEs in the industry. According to Aidexa, SME lending is historically difficult and expensive. Some of the other challenges include:

  • Small companies are more fragile in difficult environments.
  • Balance sheet data unaudited.
  • Commercial banks tend to put the most experienced people.
  • Not cost-efficient, little revenue.

On the other hand, SMEs also have significant opportunities for business growth. SMEs are the backbone of the Italian economy and the main creator of new jobs and development. 50% of Italian GDP is made up of small businesses with fewer than 50 employees

Furthermore, other aspects to consider regarding SMEs are as follows:

  • SMEs can have big growth; they are agile enough to achieve big revenue increases.
  • SMEs are owned by the founder who manages the financial strategy. In terms of client experience, SMEs behave more like a person than a company.

So, how does AideXa help address the challenges faced by SMEs?

SMEs often face difficulties when applying for loans from traditional banks. AideXa wants to help by increasing credit access to SMEs. Furthermore, the similarities in the characteristics of SMEs and individual customers drive AideXa to give entrepreneurs an immediate and simple experience as the one you can experience in a person looking for a loan.

AideXa believes that data and AI models will significantly help them in addressing the challenges faced by SMEs.

For instance, information from unaudited balance sheets may be unreliable. However, every company including SMEs possesses account transaction data, which can serve as a substitute or enhancement for balance sheet valuation.

Within account transactions, there are thousands of data points available for analyzing a company’s creditworthiness. Furthermore, to meet the transparency requirements of SME customers, AideXa has built a system that involves human interaction in credit decisions and customer relationships.

One approach employed by AideXa to fulfill these needs is the integration of Explainable AI models with a Credit Decisioning Platform, which provides transparency for clients.

AideXa uses three main pillars in utilizing data and AI models, namely:

Collecting data through a data-lake

Data is extracted from the operational system and then organized into analytical data. At this point, the data is ready to be evaluated by data scientists and business analysts. There are three layers within the data lake: bronze, silver, and gold.

Operational data is initially dumped into the bronze layer and subsequently undergoes data cleansing before progressing to the silver layer. In the gold layer, further data cleansing is performed, resulting in data of high quality equivalent to that in the Enterprise Data Warehouse (EDW). This high-quality data is accessible to all users, including Data Scientists and Business Analysts, in near real-time

For instance, when a customer completes the registration process on the website and provides various data points, the system transfers this data to the data lake within minutes.

AideXa emphasizes the importance of viewing data as a product rather than a side asset, as is often the case with traditional banks. By treating data as a product, the data stored in the data lake is ensured to be readily consumable by various stakeholders.

Consequently, AideXa claims to cut down a significant portion of the time, reducing it from the usual 10 weeks in traditional banks to just 1 week. This time saved allows the Data Science team to focus on creating Machine Learning models.

Develop Machine Learning models

The Machine Learning (ML) models use traditional data while also exploiting the granularity of account transaction data. In terms of modeling technique, use all the range of modeling techniques including Random Forest and Logistic Regression.

They use incremental learning because the data is growing exponentially.

There are five main data sources used for credit risk models. From each data source, there is at least one model:

  • Public info
    The model is based on disclosed information such as companies sector.
  • Balance Sheet
    Unaudited balance sheets are not reliable. However, revenues, earnings, and debts from the balance sheet can be used for the models.
  • External data providers
    Financial history provided by Credit Bureau (CRIF data, or SLIK in Indonesia) and Bank of Italy.
  • Transactions – aggregated
    The model is based on cash flow behavior (trends in credit and debit movements).
  • Transactions – text mining
    Information is retrieved by parsing the description of each transaction (category, counterparty, etc.).

Quick and agile runtime implementation

To achieve rapid and flexible implementation, AideXa employs two approaches, namely Banking-as-a-Service and a Credit Decision Platform.

By utilizing the Banking-as-a-Service approach, AideXa can provide a fully digital experience. This is further supported by specially designed software systems and infrastructure to operate in a cloud environment (Cloud-native components). Banking-as-a-Service also enables AideXa to create customizable APIs according to partner needs.

In addition to the use of Banking-as-a-Service, the utilization of a Credit Decisioning platform also plays a significant role in the implementation of AI within AideXa’s credit decision system.

In many cases, when discussing technology or ML models in the fintech industry, the focus tends to be primarily on the algorithms and ML models themselves. The infrastructure that facilitates the implementation of these models is usually underrated.

According to AideXa, without a solid credit decision platform, machine learning models will remain to be a theoretical concept rather than impactful elements in fintech business operations.

Hence, in AideXa, ML models are implemented in microservices while a powerful Credit Decisioning Platform orchestrates the process. The Credit Decisioning Platform is easily accessed by a set of APIs called during the onboarding process. Then, the Decisioning Platform evaluates how good the customer is in real-time, with increasing accuracy as more data is provided.

For bigger tickets, however, human intervention in credit decision-making can give better results compared to fully relying on ML models. Here, the Credit Decisioning Platform also helps organize data and provides a user interface (UI) that Credit Analysts can use to make credit decisions.

Choosing ACTICO as a Credit Decisioning Platform solution

Credit Decisioning Platform is the main actor in the runtime process as it orchestrates all the processes in the credit decision system. In addition to integration with the credit application webpage via APIs, the Decisioning Platform can provide a user interface (UI) for Credit Analysts and can also connect to the data lake.

Due to its crucial role, AideXa emphasizes that selecting the right platform is a crucial step. Therefore, AideXa utilizes ACTICO to meet their needs for a credit decision platform.

Quoting from the ACTICO website, the ACTICO Credit Decisioning Platform is a powerful software for intelligent decision automation. It combines business rules and machine learning with automation technology to make day-to-day decisions faster and smarter.

For credit decision modeling, ACTICO offers a fully graphical drag-and-drop business rules editor to design, create, and test business rules. It also enables integration with any existing ML models implemented in industry-standard languages (e.g. Python, R, SAS) and tools (e.g. H2O). Data integration is also provided for both internal and external data. ACTICO also integrates decision models into existing workflows and systems using powerful standard APIs. With the various features provided, AideXa recognizes that the flexibility and ease offered by Actico have enhanced the efficiency of their credit decision system.

In AideXa’s case, they utilize several components of ACTICO’s Credit Decisioning Platform within their credit decision system architecture. Some of the platform’s components they incorporate into their architecture include:

  • Rule Engine
    Rule engine automates decision-making processes based on business rules and ML models.
  • Underwriter UI
    This UI is used by Credit Analysts for manual decision-making when necessary.
  • Data layer
    This component integrates the Credit Decisioning Platform with the data lake.

After 3 years of business activity, AideXa has shown a proven success in their strategy. AideXa shows a high-quality customer base with a 3.4% NPE (Non-Performing Exposure) ratio of 3,400 lending customers.

AideXa’s strategies have positioned them as the institution with the fastest loan solutions in Italy, setting them apart in the banking industry. Entrepreneurs know if they are financeable in only 20 minutes. AideXa also enables entrepreneurs to receive money in their bank account in 48 hours. In comparison, typically in Italy, SMEs require a timeframe of 60-90 days for the credit evaluation process.

Banca AideXa’s approach to implementing the Credit Decisioning Platform, Machine Learning, and Open Banking has highlighted the crucial role of technology and innovation in addressing the financial needs of SMEs. With a strong focus on empowering SMEs, Banca AideXa has paved the way for more entrepreneurs to access the financing they require efficiently. Through their collaboration with ACTICO, they have created a solution that revolutionizes the process of credit decision-making.(Adri)


Actico 2023, Actico website, accessed 2 October 2023,