Build, Scale, and Govern: Recapping the EDB Postgres AI LinkedIn Webcast Series

This blog was co-authored by Purnima Phansalkar,  Jack Christie and Dunith Danushka.

Pressure to modernize has never been higher for leaders in banking, financial services, and other highly regulated industries. Data volumes are exploding, customer expectations are rising, and regulatory requirements are tightening — all while organizations are being asked to innovate faster and operate more efficiently.

This blog brings together five episodes from our Build with EDB Postgres® AI - LinkedIn Live How-to Demo Series — designed to show, not just tell, how EDB Postgres AI (EDB PG AI) can help you meet these challenges head-on.

Why It Matters

Financial institutions face a common set of obstacles: aging legacy systems, costly proprietary databases, mounting operational risk, and the need to stay compliant without slowing innovation. Many teams know what must be done — but not how to make modernization real, achievable, and secure.

Where EDB Comes In

EDB helps regulated industries modernize with confidence by providing:

• Enterprise-grade Postgres that meets strict requirements for resilience, security, compliance, and performance
• AI-ready data capabilities that accelerate insights while maintaining data control
• High availability and extreme reliability, essential for mission-critical banking workloads
• Operational efficiency through automation and cloud-native deployment options
• Cost savings and freedom from lock-in without compromising enterprise needs

These aren’t just talking points — they are capabilities our customers use every day to reduce risk, modernize applications, and build secure, scalable, future-ready data foundations.

From Talking to Showing

It’s easy to speak in big ideas and bold promises. But how do these solutions come together in practice? That’s exactly why we created this live demo series — short, high-impact episodes that walk through real patterns, real architecture, and real implementation examples teams can use today.

Watch the Recaps

If you missed the live sessions — or want to revisit a specific topic — here are the full replays of all five episodes. Each one is designed to give you practical insights and takeaways for your data modernization journey.

Episode 1: Chat Assistant as an Internal Knowledge Base 

• The Challenge: Frontline workers struggle to quickly access accurate and compliant information scattered across disparate systems, including unstructured data residing in places like object stores and also structured data locked up in databases.
• The Solution: This session demonstrates how to implement a virtual assistant using EDB PG AI Factory. Leveraging our GenAI Builder, AI Pipelines, and a vector Knowledge Base, the assistant can quickly turn institutional knowledge (long-term memory) into factual, compliant answers grounded in data. The session also showcases how the Assistant’s responses are grounded in both structured and unstructured data and constrained with proper guardrails.

• Watch the replay and demo HERE

Episode 2: Querying and Governing Lakehouse Tables 

• The Challenge: Organizations accumulate massive volumes (terabytes, possibly petabytes) of event data—such as transactional data, and digital activity—in an S3-based lakehouse (Iceberg/Parquet files). They need fast, scalable, and SQL-friendly query interface to generate Customer 360 insights.
• The Solution: This episode showcases how to use the EDB PG AI Analytics Accelerator and the embedded Iceberg Catalog to query and govern Iceberg tables directly from Postgres. The Analytics Accelerator acts as a Postgres frontend to the lakehouse ecosystem, allowing familiar Postgres tools (like PGAdmin) to query the data without moving or loading it into the database. This enables real-time queries for customer segmentation and outreach campaigns.

• Watch the replay and demo HERE

Episode 3: Multi-Modal Search Across Diverse Data 

• The Challenge: Detecting insurance fraud requires processing and searching unstructured and visual data, such as crash images, efficiently. Traditional keyword search is insufficient for analyzing these visual assets.
• The Solution: The session demonstrates building a search engine capable of querying structured, unstructured, and semi-structured data using low-code AI Pipelines in EDB PG AI Factory. By using declarative SQL and the CLIP model, the pipeline extracts images (from sources like S3 or MinIO for self-hosted sovereignty) and generates vector embeddings, which are stored in Postgres. This process enables advanced semantic search, allowing users to search visual content (like accident images) using natural language queries.

• Watch the replay and demo HERE

Episode 4: Agentic Analytics with AI Agents 

• The Challenge: Account executives (AEs) need to instantly combine customer engagement data scattered across multiple silos — such as transaction history, loyalty status, and product eligibility documents — to uncover upsell and cross-sell opportunities. Today, they rely on IT or analytics teams to pull this data together, which slows down decisions and impacts revenue moments.
• The Solution: This episode demonstrates how to combine EDB PG AI Analytics Accelerator and AI Factory in an agentic analytics solution, delivering real-time, trusted insights directly to AEs and other business users. Analytics Accelerator—powered by its Lakehouse Connector and Analytics Engine—provides seamless access to both Postgres CRM/loyalty data and lakehouse transactional datasets through a unified analytics layer. AI Factory builds on this foundation with an enterprise-ready AI agent equipped with a reasoning model, long-term knowledge base, and Python tools created in Agent Studio. Supported by AI Pipelines, the agent orchestrates data retrieval, interprets context, and generates personalized, compliant recommendations with full traceability. AEs get accurate, in-the-moment insights without waiting on backend teams, enabling faster, smarter revenue decisions.

• Watch the replay and demo HERE

Episode 5: Working with Time-Series Data in Postgres 

• The Challenge: Financial institutions process millions of append-only credit card transactions daily. They struggle with database bloat, high storage footprint and meeting strict data retention policies mandated by regulators.
• The Solution: The episode introduces the Bluefin extension, a Postgres table access method designed for time-series workloads. Bluefin uses delta-compressed storage, resulting in a significantly reduced storage footprint compared to standard heap tables. The append-only nature of the data also allows for faster query performance for trend analysis and aggregations. This enables analysts to sustain rapid, real-time dashboarding and trend analysis on massive data volumes directly from Postgres.
• Watch the replay and demo HERE

Ready to Build Your AI and Analytics Future?

Whether you’re modernizing transactional systems, unifying analytics across your lakehouse, or moving toward autonomous, agent-driven workflows, EDB Postgres AI gives you a sovereign, end-to-end foundation to build confidently — from your core operational databases all the way to enterprise-grade AI agents.

With one platform that spans ingestion, governance, analytics, vector search, pipelines, knowledge bases, and agent orchestration, EDB Postgres AI lets you innovate on your terms while keeping control of your data, your cost structure, and your compliance posture.

Take the Next Step

Sign up for a free, one-on-one workshop tailored to your specific use case. Partner directly with EDB experts to scope your requirements and receive a customized, hands-on session designed to accelerate your path from concept to production. Book your free workshop today!