The AI economy is projected to reach $17 trillion by 2028, fundamentally altering how organizations architect their infrastructure. Driven by this shift, 95% of major global enterprises are on a mission-critical sprint to become their own AI and data platforms within the next two years. 

Yet only 13% of enterprises have successfully found the formula. Their secret to mainstreaming agentic AI? Abandoning fragmented, legacy architectures and placing their data directly alongside their AI in a secure, compliant, and sovereign manner.

As organizations rapidly transition to an “agentic” workforce, they are entering a highly volatile, uncertain, complex, and ambiguous (VUCA) environment. Surviving this shift requires abandoning rigid, traditional strategies in favor of agility and resilience. For enterprises leading the charge, the foundational layer of choice is clear: true open source relational databases. Today, 81% of these successful enterprises have committed to open source strategies, with over 40% standardizing on PostgreSQL as their relational data layer.

As Doug Flora, VP of Product Marketing at EnterpriseDB (EDB), noted: “It’s imperative in moments of rapid change to follow the patterns of the leaders looking to forge success, not the majority who are still working in the patterns of the recent past. Those committing to open source and a mission-critical focus on sovereignty over their AI and data are plotting a pathway to agentic success that achieves 5x the ROI of the majority.”

Extensibility matters: AI needs both structured and unstructured data

AI applications cannot run on vector embeddings alone; they require a deep synthesis of structured, semi-structured, and unstructured data. Unlike many legacy databases that bolt on new features as afterthoughts, Postgres was natively architected for core extensibility. It empowers developers to extend data types, indexes, query planners, functions, and storage engines dynamically. 

By unifying vectorized data with traditional transactional (binary) data, Postgres effectively gives AI agents the “eyes, ears, and brain” necessary to sense inputs and operate autonomously within a single, ACID-compliant environment.

An ecosystem built for architectural agility 

In a rapidly expanding data ecosystem, relying on a fragmented architecture of specialized databases creates complex synaptic connectors prone to latency, integration failures, and data silos—or what amounts to human hallucinations at the system level. Postgres eliminates this technical debt by extending a single database engine to meet diverse workload demands.

“Developers have long loved Postgres for its extensibility, flexibility, and open innovation model. Now global enterprises are recognizing that same value, making Postgres a strategic decision and running mission-critical data systems on it,” said Jozef de Vries, SVP, Core Database Engineering, EDB.

Developers can seamlessly extend Postgres to handle highly complex, volatile workloads:

• pgvector: Enables advanced vector search, allowing developers to combine relational data, metadata, and embeddings to build robust retrieval-augmented generation (RAG) applications
• Citus: Accelerates multi-tenant SaaS applications and powers real-time analytics (HTAP) via transparent sharding and parallel query execution
• PostGIS: Delivers enterprise-grade geospatial querying, critical for defense and retail industries
• TimescaleDB: Manages massive time-series data crucial for complex analytic models and agentic learning patterns
• pgraph: Handles complex, interconnected data traversals to uncover hidden relationships

The future needs crowdsourced intelligence, not vendor lock-in

Crucially, no single corporate entity owns Postgres. Its vitality relies on the collective intelligence of one of the largest independent developer communities on the planet. In 2025 alone, more than 260 developers contributed code directly to PostgreSQL’s core database engine, with hundreds more participating in testing, reviews, and documentation across the world. Beyond the codebase, the ecosystem is supported by hundreds of user groups, meetups, and international PostgreSQL conferences that keep innovation flowing across all five continents.

While enterprise-grade platforms are built around Postgres to optimize it for sovereign, agentic environments—with big tech giants among the top commercial contributors and EDB leading with more than 30% of contributions—its innovation comes directly from this rich and diverse community that continues to expand. Drawing on the principles of James Surowiecki’s The Wisdom of Crowds, this crowdsourced intelligence ensures that the database evolves faster and more robustly than it would in any proprietary, single-vendor ecosystem.

Securing a sovereign data future

To thrive in the agentic future, engineering and data leaders must make two critical architectural moves: First, break free from locked-in legacy relational ecosystems, such as Oracle, MySQL, SQL Server, or Greenplum , that constrain agility.

Second, harness the immense extensibility of Postgres, its vibrant open source community, and its core ACID capabilities to unify data and AI.

The future of enterprise architecture isn’t about renting space in a hyperscaler’s proprietary ecosystem. It’s about creating your own sovereign platform, where your structured and unstructured data seamlessly power a new agentic workforce under your complete control. Move your data to Postgres now, or risk missing the foundation of the agentic future.

Get your complimentary copy of the O’Reilly book Building a Data and AI Platform with PostgreSQL.