Zoral Data Fabric
Zoral Data Fabric is the memory of the Zoral Finance Operating System (fOS). It records every application, decision and change as it happens, in a banking data model with quality checks built in. Your credit team, auditors and supervisors work from one record. Your core stays as it is.
At the centre of Zoral fOS, Zoral Data Fabric remembers, Zoral Decision Engine decides and the Business Outcomes Layer measures. Zoral Integration Gateway brings in data from your systems. Every workflow step and every decision, with its reasons, is recorded here.
The foundation of your enterprise data lake. Because it records every application, decision and change across Zoral fOS and its solutions, it becomes one of the main trusted sources for your enterprise data lake. Atelier’s Data, AI & Automation team extends that lake with data from your core, cards, CRM, finance and partners. Scores and decisions flow back into Zoral Data Fabric and on to the lake. Every team works from the same record.
Activity on Zoral fOS is recorded as it happens. Reports and decisions do not wait for the overnight batch.
Every decision is kept with its reasons. Your teams spend less time gathering evidence for supervisors.
Your data scientists tune machine-learning models on this record. The models run in Zoral Decision Engine, and their scores and decisions are written back here.
Each status change, document update and calculated field is recorded, in every channel.
Each status change, document update and calculated field is recorded, in every channel.
Related changes are saved completely or not at all. An application update and the bureau response that came with it are saved as one unit, so a decision does not work from half a picture. ACID (atomicity, consistency, isolation, durability) applies to writes into Zoral Data Fabric.
Changes are added as new time-stamped entries, not overwritten. Point-in-time queries show the data at any past moment.
A documented relational banking data model, with primary and foreign keys, which your team extends.
Relational stores such as PostgreSQL, plus NoSQL and in-memory stores. Access by API, SQL, GraphQL, events and MCP, under role-based permissions.
Your data scientists’ own machine-learning platform, such as Amazon SageMaker, Databricks or Google Cloud Vertex AI, reads the data and sends scores back.
Guardrailed, with human-in-the-loop supervision. Agents prepare, check and draft, only within the permissions, workflows and approvals your bank sets. Agents do not make credit decisions. Those follow rules and models your bank approves, and accountability stays with your bank. Your bank chooses its language models, connected through MCP, and data stays within the environments and agreements it controls. Every action, prompt and tool call is recorded.
AWS or Azure in your chosen region, or your own data centre.
| What you compare | Zoral Data Fabric | Data-virtualisation tools | General integration hubs |
|---|---|---|---|
| How it works | Records each step as it happens. | Queries your databases on demand. | Copies records between systems, often on a schedule. |
| Consistency | Related data is saved together or not at all. | Depends on each underlying database. | Systems can disagree until they catch up. |
| Banking data model | Included, ready to extend. | None. Your team designs it. | None. Your team maps each field. |
| Load on your core | Analysis runs on the record. | Queries can slow an older core. | Can add load, depending on how each link is built. |
| Effort to connect | Configured in Zoral Integration Gateway. | Each view is mapped by hand. | Your team builds and maintains each link. |
Zoral Data Fabric
Records each step as it happens.
Data-virtualisation tools
Queries your databases on demand.
General integration hubs
Copies records between systems, often on a schedule.
Zoral Data Fabric
Related data is saved together or not at all.
Data-virtualisation tools
Depends on each underlying database.
General integration hubs
Systems can disagree until they catch up.
Zoral Data Fabric
Included, ready to extend.
Data-virtualisation tools
None. Your team designs it.
General integration hubs
None. Your team maps each field.
Zoral Data Fabric
Analysis runs on the record.
Data-virtualisation tools
Queries can slow an older core.
General integration hubs
Can add load, depending on how each link is built.
Zoral Data Fabric
Configured in Zoral Integration Gateway.
Data-virtualisation tools
Each view is mapped by hand.
General integration hubs
Your team builds and maintains each link.
General data tools do not know what a loan balance, a product or a regulatory field is. Here, they are already defined.
Virtualisation tools can query your core while customers apply. Here, reports run on the record instead.
When data is copied later, credit and fraud checks can miss the customer’s latest step. Here, each data point is saved when it is created.
Accounts, channel activity, internal risk measures and bureau histories sit in one model.
Your data scientists retrain models on clean, current history. Atelier’s Data, AI & Automation team can work with them.
Reports draw on the same time-stamped record that decisions use.
Applications are checked against known fraud patterns while in progress. Zoral Behavior Matrix, a separate Zoral product, can add behavioural signals where your bank switches it on.
While the customer is still applying, Zoral Decision Engine can offer a better-matched product or price, within your credit policy.
The Challenge
Separate databases and overnight batches left risk reports out of date.
The Solution
Zoral Data Fabric now records product actions and transactions as they happen.
The Strategic Outcome
The bank tracks key portfolio indicators on current figures.
The Challenge
Missing data on business borrowers led to inconsistent scores and too many manual reviews.
The Solution
We linked origination, underwriting and portfolio data in Zoral Data Fabric.
The Strategic Outcome
Scores became more consistent, defaults fell and more good applicants were approved.
Yes. Each change is added as a new, time-stamped entry. Your team can look at any past moment and see the inputs, policy version and model version used. This supports fair-lending reviews and supervisory requests.
They could. But general tools do not know what a loan, a limit or an arrears status is. Your engineers would build that banking data model and maintain it for years, for every product you add. Zoral Data Fabric comes with one, ready to extend.
Lineage is the other reason. When each step runs in a different tool, the trail from source data to decision breaks at every hand-off. On Zoral fOS, lineage, data-quality checks and the audit trail come built in. Your engineers keep full access to the data, by API, SQL or events.
No. It feeds it, as one of the main trusted sources. Your lake stays on your own platform, such as Databricks or Google Cloud BigQuery. Atelier’s Data, AI & Automation team connects the two, with data flowing both ways.
Atelier, Zoral’s senior consulting practice, as part of each Zoral fOS programme. Data, AI & Automation links it into your data lake and data-quality work. Engineering and delivery handles the technical set-up.
Product brief
9 pages to read and share with your colleagues: what it changes, how it is controlled and audited, and how it works with the tools you already own.
See how Zoral Data Fabric records a decision, and how your team reproduces it months later.
Prefer to pick a time? Schedule a meeting
Automate lending, banking and finance operations.
London
Miami
We are on Socials
+44 20 3963 2002