Oracle Developers

Congrats to the 2026–2027 Oracle ACE Program members! 🥳 Learn more about this year's members and how their passion for sharing knowledge, supporting the community, and helping others succeed continues to make a difference across the Oracle ecosystem. social.ora.cl/6013BDcjcB

Building AI agents that actually remember? 🧠 The latest Oracle AI Agent Memory update adds: ✅ Custom memory extraction ✅ Hybrid search (semantic + structured retrieval) ✅ More control over what agents remember and retrieve ✅ Better governance for production AI These capabilities help improve retrieval quality, reduce noise, and give you finer control over long-term memory—all while keeping memory governed inside Oracle AI Database. Read more: social.ora.cl/6014BEKoBY

Watch this introduction to the image generation models available on OCI Enterprise AI, including an end-to-end walkthrough of the Images API and live demo on integrating image generation into an application. social.ora.cl/6012BDJ2Gn

Oracle Select AI for Python enables you to use Oracle Select AI to build Python applications, notebooks, services, and command-line workflows. With Select AI for Python 1.4, the focus shifts toward application development: 🗣️ Streaming chat and streaming generation 🐍 Python web framework patterns 💻 Concurrent prompt processing Learn more: social.ora.cl/6011BDJFDz

Organizations are generating and retaining more data than ever before. The challenge now is not simply where to store cold data, but to ensure that archived data remains accessible, recoverable, secure, and cost-effective over time. See how OCI provides a strong enterprise cloud foundation for managing data across multiple storage tiers. And when paired with Geyser Data, organizations can build a modern cold data archiving strategy that reduces long-term storage costs while maintaining confidence in data access and recovery. social.ora.cl/6013BDJxRV

Share your leadership and expertise with other organizations at #AIWorld. Apply to speak by July 20, and if selected, you'll receive a complimentary pass. social.ora.cl/6016BEHaYo

Startups, start your engines! We're proud to be the Founding Partner of IMSA Labs, a platform for innovation and collaboration across motorsports, automotive, and technology. As part of this partnership, we’re launching Oracle Cloud Innovation Studio—the first initiative within IMSA Labs designed to help startups by combining our cloud and AI technologies with @IMSA’s live race operations. Learn more: social.ora.cl/6011BEKGbb

Part 2 of our series about using a GPU to offload the creation of vector indexes on Oracle AI Database 26ai covers configuration of the Vector Index Service on Oracle AI Database 23.26.2. social.ora.cl/6018BDGJLq

It’s the final week to save $700 on your #AIWorld pass! social.ora.cl/6016BEDer6

Six weeks ago we released @Oracle AI Agent Memory. A Python package for giving agents durable memory on @OracleDatabase. Then we did the part that actually matters. We listened. Customers and the developer community told us where it helped, where it hurt, and what real agents need once they leave the demo and hit production. One theme kept surfacing. Tokenomics This is the word that you will probably hear for the rest of the year. It basically means how can I spend less on my model inference, with the caveat of not have degrading quality. Tokenmaxxing was great while it lasted But every token you put in context has a price. Latency, cost, and the model's attention all scale with it. Agent memory is the discipline that makes agent economics work at all. So over the last month, the team rebuilt around exactly that. Here is what the team shipped in the latest release of the Oracle AI Agent Memory Package⤵ 1️⃣ Background extraction. The agent responds now. Turning a messy conversation into durable memory happens in the background, so nobody is left waiting on it. 2️⃣ We added retrieval controls so context stays within budget. You set the caps: how many memories come back, how many recent messages replay, and the token budget for the generated summary. 3️⃣ Hybrid search. Vector search is not all you need. 4️⃣ Chunked indexing: Long documents are indexed as chunks but returned as one logical result, so retrieval stays sharp on a big policy. 5️⃣ Context cards: This is tokenomics in action, we provide every agent loop a compact, prompt-ready card replaces replaying the full transcript. Less context, lower cost, sharper focus. 6️⃣ Custom extraction instructions. Tell the extractor to keep the durable facts and drop the greetings, secrets, and throwaway wording. 7️⃣ Typed memories. Facts, preferences, guidelines, and general recollections stay distinct, because an agent should treat them differently. 8️⃣ Exact scope matching. A semantically perfect result is still wrong if it belongs to another customer. exact_user_match=True stops memory crossing that line. And there's plenty more All the new features can be explored in a single notebook The end-to-end support copilot notebook (the fastest way to see it work) github.com/oracle-devrel/… We are just getting started.

How can you skip elements in a stream? @JosePaumard breaks it down.

Go behind the scenes with us as we explore how OCI Secret Management Service designed cross-region replication to help customers strengthen disaster recovery for critical secrets. social.ora.cl/6010BD0P2u ➡️ Learn how the team used asynchronous replication, regional isolation, API versioning, and sequence numbers to reduce blast radius and protect reliability across outages. ➡️ See how thoughtful distributed systems design helps keep secrets available when it matters most.

You asked. We listened. @thatjeffsmith shares details about Color Coded connections in SQLDev for VS Code. social.ora.cl/6010BEEDGA

Spec-Driven Development with Oracle APEX blueprint scaffolding provides a structured, repeatable path from business intent to working application. See how it accelerates proof of concept, reduces token consumption, and improves generation quality by constraining the AI to a well-defined intermediate format. social.ora.cl/6011BDHEN7 #orclAPEX

The Oracle AI Accelerator Pack for Vehicle Route Optimizer uses NVIDIA cuOpt open-source GPU-accelerated optimization engine on OCI GPU infrastructure to help customers move large routing problems from overnight planning cycles to minute-scale optimization. Learn more: social.ora.cl/6016B8Lpja

With Oracle AI Database Private Agent Factory and Exadata Database Service for Developer, we're offering a practical path for teams to build, test, and validate enterprise-grade agentic applications on Exadata infrastructure without Oracle Database license costs for development and testing. Learn more: social.ora.cl/6010BDGJRq

Want to achieve sub-millisecond latency and highly consistent database performance for your Oracle workloads on AWS? Then Oracle Database@AWS high performance networking is a feature you should explore! ⚡ We show you how Oracle Database@AWS high performance networking works, its key benefits, and basic configuration steps. social.ora.cl/6019BDJq8t

Building AI apps is easy. Building AI apps that only access the data users are authorized to see is a different story. We show you how to integrate Oracle Deep Data Security with Spring Boot and Spring AI to enforce identity-aware, fine-grained authorization directly in the database. Instead of relying on prompts, application logic, or LLM guardrails, access control is enforced where the data lives. social.ora.cl/6010BDJUry

Soccer analytics is becoming increasingly sophisticated, helping teams and fans use data to understand performance, tactics, and player impact. ⚽ AI agents allow you to go a step further. Watch this code-along webinar from @DataCamp with our own Ignacio (Nacho) Martínez Rincón as he guides you through building a soccer analytics agent using Claude and Oracle as the memory layer. social.ora.cl/6016BDcQ6M

APEXlang Atlas helps you navigate and understand the world of Oracle APEX components and properties. We show you how Its searchable trees of Components and Properties let you discover interesting Details about your current selection. By setting property values, picking from LOVs where available, you can see which child components and other properties become relevant or are no longer applicable. social.ora.cl/6014BDHBMI #orclAPEX