Connect every database. Move zero data.

No. Fyno Connect runs as a Docker or Kubernetes container deployed inside your own cloud or on-premise infrastructure. It establishes read-only connections to your databases and data warehouses. Queries execute against the source database and return only the result set needed for the current operation. Customer data is never replicated, exported, or stored outside your infrastructure. You can inspect network traffic from the Fyno Connect container and confirm that no customer records are transmitted externally.

Oracle, SQL Server, PostgreSQL, MongoDB, Snowflake, BigQuery, Redshift, Athena, and S3. The connector supports both relational databases and cloud data warehouses. For banks running core banking systems on legacy databases, Fyno Connect reads customer data in real time: loan amounts, EMI due dates, account statuses. No data exports, CSV uploads, or ETL pipelines required.

Fyno Connect supports both approaches simultaneously. Kafka-based event streaming provides lower latency: your core banking system writes events to a Kafka queue, and Fyno Connect picks from the queue and triggers messages immediately. Database polling checks for new records at configurable intervals. Most banks use Kafka for real-time events like OTPs and transaction alerts, and database polling for batch operations like scheduled campaign sends. You can run both in parallel.

Full on-premise deployment. The Fyno Connect container runs on your own Kubernetes cluster, OpenShift, or Docker environment within your AWS, Azure, or private cloud. Fyno's team deploys and maintains the containers while you manage the infrastructure. All processing happens within your security perimeter. An alternative hosted model runs on AWS Mumbai region for organisations that accept managed cloud deployment. Protium, Shriram Finance, and Babalease use the private cloud model.

Fyno's AI reads your database schema (table and column structure, not customer records) to translate natural language segment definitions into queries. Only structural metadata is shared with the AI engine. The generated query runs against your connected database through the on-premise connector entirely within your infrastructure. No customer data, including result sets or customer IDs, is sent to Fyno's servers. Marketing teams define audiences using descriptions like "customers with EMI due in 3 days" without writing SQL. Cohorts refresh dynamically: whenever a campaign runs, the query executes against the latest data in your database, so that the audience is always current at the moment of dispatch.

Before every communication send, Fyno Connect reads the latest consent status from your CRM or database. This ensures that preference changes captured in your own systems are respected by Fyno's routing engine without requiring real-time API calls or webhook integrations. DND status is validated at the TRAI registry level before any promotional send. The sync covers multi-channel consent and topic-level preferences.

Fyno Connect runs as a lightweight container on any Kubernetes cluster or Docker runtime, whether cloud-hosted or on-premise. A Redis instance is required for operational data: your central Redis works, or Redis can run within the same Docker setup. Network access is needed to your source databases and to Fyno's orchestration layer for campaign triggers. Fyno helps with initial deployment and publishes application patch updates to Docker Hub for your team to pull. Fyno maintains the application code behind Connect. OS-level patching and infrastructure maintenance remain with your team. Resource requirements are modest: the container handles query execution, not data storage beyond the Redis layer.