How We're Bringing Postgres Into the AI Era

Hien Phan -

No matter what the latest computing requirements bring—from real-time analytics to AI—Timescale is here to help developers unlock the full potential of Postgres. We believe that every developer should be able to create state-of-the-art applications on the database they already know and trust without extra infrastructure or specialized expertise.

We started by extending Postgres for real-time analytics and time-series workloads. Today, thousands of developers across industries—including IoT, manufacturing, energy, finance, and more—rely on Timescale for seamless, data-intensive applications. Now, as AI reshapes application development, we’re taking Postgres even further, enabling developers to bring AI capabilities directly into their database.

October marked a landmark shift: we transformed Postgres into a full AI development platform, modernized its GUI with AI, and enhanced usability and performance for real-time analytics. Dive in and see how these new tools can help you build faster, smarter, and with fewer obstacles.

Pgai Vectorizer: Automate AI Embeddings With One SQL Command in Postgres

Creating embeddings is as simple as creating an index. Building agents, chatbots, and recommendation engines shouldn’t require complex, specialized infrastructure. Pgai Vectorizer brings vector capabilities directly into Postgres, automating embedding workflows and keeping embeddings in sync with your source data, just like indexes.

With pgai Vectorizer, you can:

  1. Generate and store embeddings with SQL: create vector embeddings from text directly in Postgres.
  2. Stay synchronized: embeddings automatically update as data changes, so you’re always up-to-date.
  3. Experiment quickly: test and switch AI models with SQL commands—no reprocessing needed.

Here’s how to generate embeddings in Timescale:

-- Create and store vector embeddings in Postgres
SELECT ai.create_vectorizer(
   'public.blogs'::regclass,
   embedding => ai.embedding_openai('text-embedding-3-small', 1536),
   chunking => ai.chunking_recursive_character_text_splitter('content'),
   formatting => ai.formatting_python_template('Title: $title\nURL: $url\nContent: $chunk')
);

Built to scale. As your embeddings grow, the pgai suite scales with you, automatically optimizing search performance using high-efficiency indexes like HNSW or StreamingDiskANN once your vector count exceeds 100,000.

Try pgai Vectorizer free in Early Access on GitHub or Timescale Cloud and build your next AI application using familiar SQL.

SQL Assistant: Code Faster and Smarter With an AI Companion

Smart, context-aware SQL support, built right into Timescale Console. Writing SQL can be powerful, but even the best developers hit roadblocks. SQL Assistant is here to help—providing real-time support for SQL writing, debugging, and optimization directly in Timescale Console.

Unlike general-purpose large language models (LLMs), SQL Assistant is fully integrated with your database, using schema-aware, context-specific insights to provide help where it’s needed most. It’s not just an assistant; it’s a smart SQL partner that adapts to your workflow.

Here’s what SQL Assistant can do:

  • Help you in real time: No more copying and pasting between tools. Access instant guidance on functions like LAG() or troubleshoot complex queries right in your editor.
  • Troubleshoot instantly: SQL Assistant flags errors and suggests fixes as you code, so you stay focused.
  • Keep your work organized: Auto-generated titles and summaries help you keep your query library clean, clear, and easy to share.

If you’re building an application and hit a syntax error, SQL Assistant has you covered. No more jumping between your editor and documentation—SQL Assistant helps you troubleshoot and optimize on the spot.

Now in Early Access on Timescale Console, SQL Assistant is free to use and privacy-first. All AI features are off by default, activated only by Timescale Cloud project members or PopSQL admins, ensuring your data remains private and secure.

Try SQL Assistant, give us feedback, and let’s shape the future of the Postgres GUI using AI together.

Foreign Data Wrappers: Connect Distributed Data Without Extra Steps

Unified data access across distributed systems. Imagine you’re part of an e-commerce team that started with all data in Amazon RDS. As your application’s real-time analytics needs grew, you adopted Timescale to handle time-series metrics and other analytics data while keeping transactional data, like customer orders, in RDS. 

Now, you need to perform analysis that combines data from both Timescale and RDS. Instead of building a complex pipeline to consolidate data, you can configure Foreign Data Wrappers (FDWs) in Timescale Cloud to seamlessly query data from external sources—whether other Timescale Cloud services or PostgreSQL databases outside of Timescale.

FDWs also support VPC peering, allowing queries from your Customer VPC to a Timescale VPC service within the same project (though not from Timescale VPC to Customer VPC). This is especially useful if you manage multiple Timescale Cloud service types and need to integrate time-series and regular data.

For more details on setting up FDWs, check out our documentation.

Chunk Size Recommendations: Optimizing for Performance

Automated recommendations for peak performance. Real-time analytics relies on optimal configuration, and even small settings—like chunk size—can impact performance. If chunks are too small, queries may slow down due to metadata bloat and extended planning times. Chunk Size Recommendations uses platform insights to analyze your database and suggest optimal chunk sizes for your hypertables, ensuring efficient queries without requiring expert DBA (database administrator) skills.

If Timescale detects that a table’s chunks are undersized, you’ll see an alert with recommended adjustments directly in the Timescale Console. You can follow the recommendations step-by-step, from identifying impacted hypertables to adjusting chunk sizes. Implementing these recommendations improves database performance as your data grows.

Later this week, we’re also launching EXPLAIN ANALYZE, a video series where our PMs share practical insights and best practices for building fast, efficient applications on Timescale. Whether you’re refining a PoC or scaling to production, EXPLAIN ANALYZE offers quick, actionable videos to help you get the most from your database. Stay tuned for more!

Open Access to Powerful Tools for Every Developer

At Timescale, our mission is to enable you to build state-of-the-art applications—all on Postgres, without needing new infrastructure or specialized expertise. As each computing era brings new demands, like AI’s rapid impact on software development, Timescale ensures you have the tools to keep pace, whatever the future holds.

With pgai Vectorizer for AI embeddings, any developer can unlock AI capabilities directly within Postgres. SQL Assistant serves as your intelligent SQL companion, helping you work smarter, not longer. Foreign Data Wrappers enable seamless access to distributed data, while Chunk Size Recommendations optimize performance—transforming Postgres into a flexible, powerful platform tailored for your workflow.

We’d love to see how these tools support your projects and hear how we can continue improving Postgres for you. Explore our open-source work on GitHub, try Timescale Cloud, and share your ideas with us. Let’s keep pushing Postgres forward, together.