Overview
The Redshift integration is used to make warehouse-resident customer and product data available to Cora.ai for analytics, enrichment, and downstream workflows.
Important: This integration is intentionally discussion-led. The exact setup depends on your Redshift schema, data volume, update cadence, and security requirements. We will align with your data and IT teams before finalizing the approach.
Recommended Approach: Export data from Redshift to Cora.ai via SFTP rather than providing Cora.ai with direct database access.
This keeps Redshift isolated within your network while allowing controlled, auditable data exchange.
Why We Recommend Redshift → SFTP (Instead of Direct Access)
Using SFTP as the handoff mechanism provides several advantages:
Strong security boundary
No direct database credentials, VPC peering, or inbound network access required
Read-only, file-based data sharing
Predictable governance
Explicit control over which datasets and fields are exported
Easier auditing and change management
Operational simplicity
Compatible with existing batch pipelines and data ops workflows
Avoids coupling Cora.ai to your warehouse schema or query patterns
Clear ownership model
Your team owns data preparation and export
Cora.ai consumes only what is intentionally shared
If your use case requires near-real-time access or ad-hoc querying of large datasets, a direct Redshift integration can be evaluated as a follow-up option.
What We Will Align on During the Setup Discussion
In a working session with your data and IT owners, we will align on:
Which Redshift datasets are in scope (accounts, usage metrics, events, health scores, etc.)
File format and structure (CSV, Parquet, JSON, compression)
Export cadence (daily, intra-day, or on-demand)
Delivery guarantees and retry behavior
PII handling and data minimization expectations
Validation and reconciliation approach
Implementation Path
Your team is responsible for generating and delivering files to an agreed-upon SFTP endpoint.
High-level steps:
Define export views in Redshift
Create stable views or queries that represent the datasets to be shared.
Avoid exposing raw or intermediate tables unless required.
Generate export files
Produce files on a scheduled basis.
Include deterministic identifiers for idempotent processing.
Deliver files via SFTP
Upload files to the agreed directory structure.
Follow naming conventions agreed during setup (for example: dataset name, date, version).
Cora.ai ingestion and validation
Cora.ai ingests files, validates schema and row counts, and surfaces any errors.
Failed or partial ingestions are reported back for remediation.
Downstream usage in Cora.ai
Successfully ingested data is mapped into Cora.ai workflows, agents, and reporting surfaces.
Notes and Guardrails
Discussion first: The exact schema and cadence are finalized collaboratively.
Data minimization: Only export fields required for operational use.
PII: Sensitive fields should be explicitly reviewed and approved.
Ownership: Your team owns Redshift exports; Cora.ai owns ingestion and downstream processing.