Uploaded May 2026 | Updated September 2026, 2 weeks ago
We spoke with @itsthatladydev on where developers can learn more about AI at Oracle Creator Labs in San Fran this week
We spoke with @itsthatladydev on where developers can learn more about AI at Oracle Creator Labs in San Fran this week

#springboot #springai #oracle #ai #rag #java Make your AI Chatbot Remember Conversations (Spring Boot + Oracle)](https://i.ytimg.com/vi/Vh5nxDeiRFs/mqdefault.jpg)



![Spatial AI Thinking in theOracle AI Database:An Introduction to Spatial AI Skills
Join the Oracle Autonomous AI Database and Spatial PM teams in this Learning Lounge session on using documentation-backed AI Skills to generate trustworthy Oracle Spatial SQL. Learn how Oracle Database Skills help AI assistants move beyond plausible-looking answers by grounding them in curated Oracle guidance, examples, version notes, and common mistake patterns. We¿ll demonstrate how spatial queries expose the difference between guessed SQL and correct, index-aware Oracle SQL.Ask any AI assistant to Find the three nearest stores to this customer, and it will hand you SQL that looks right. Against an Oracle database, its often subtly wrong ¿ the wrong operator, latitude and longitude swapped, the spatial index quietly bypassed so the query scans every row. The gap isnt the models reasoning; its that the model is guessing instead of following Oracles own documentation. Oracle Database Skills close that gap ¿ and spatial data is the perfect place to see why.What are Skills?Oracle DB Skills is a library of 100+ documentation-backed guides for working with Oracle Database ¿ but theyre not blog posts for humans to skim. Each Skill is curated, source-cited expertise packaged as context an AI assistant loads before it answers: best practices, real examples, version notes for 19c and 26ai, and a hard-won catalogue of common mistakes ¿ all engineered to produce correct Oracle code instead of plausible-looking code. Well show what a Skill actually contains and how an AI tool consumes it.
Why spatial is the killer demoSpatial querying is non-obvious and unforgiving. The SDO_GEOMETRY object type has five parameters and encodes the geometry type as four-digit codes; you must register metadata before an index can be built; and the traps are silent and expensive ¿ put latitude before longitude and your point lands in the ocean, or call a measurement function in a WHERE clause and your index never fires. These are exactly the mistakes an ungrounded model walks straight into, which makes the before-and-after impossible to ignore.
What youll see
A live before-and-after: the same natural-language request answered without a Skill, then with one ¿ wrong-but-confident SQL versus correct, index-aware Oracle Spatial SQL.
The spatial essentials the Skill encodes ¿ SDO_GEOMETRY, R-tree indexing and why metadata comes first, and choosing between SDO_WITHIN_DISTANCE, SDO_NN, and SDO_RELATE.
How the SQLcl MCP server connects your AI assistant to a live database, so it can inspect schemas, run the query, see the error, and correct itself ¿ a closed loop from English sentence to validated result.
Youll leave with a clear picture of what Skills are and why they work, a feel for Oracles spatial building blocks, and a repeatable pattern for getting trustworthy SQL out of an AI assistant.
Outline
05:11 - The goal of this session
04:21 - What are AI Skills
20:06 - Adding Skills to your AI Tools
30:28 - Closing the loop with the SQLcl MCP Server
37:15 - Practical use of Spatial Skills
56:17 - Key takeaways
Objectives
By the end of the session, you will be able to:
Explain what Oracle Database Skills are and how an AI assistant uses them as grounding context before answering.
Recognise why ungrounded language models produce subtly wrong Oracle SQL ¿ wrong operator, swapped coordinates, bypassed index.
Identify the core Oracle spatial building blocks ¿ SDO_GEOMETRY, R-tree indexing, and why metadata must come first.
Choose the right spatial operator for a task among SDO_WITHIN_DISTANCE, SDO_NN, and SDO_RELATE.
Describe how the SQLcl MCP server connects an assistant to a live database to form a closed loop from question to validated result.
Apply a repeatable pattern for getting trustworthy SQL out of an AI assistant in their own work.
Useful Links
Agent Skills Overview
Agents Skills: load CSV to new Oracle tables via SQLcl [ MCP by Jeff Smith|https://www.thatjeffsmith.com/archive/2026/03/agents-skills-load-csv-to-new-oracle-tables-via-sqlcl-mcp]
Having a go with 100+ new AI Skills for Oracle AI Database by Jeff Smith
Oracle Skills GitHub repo - App Dev specific skills
AI For You Newsletter by Kay Malcom on LinkedIn
AI agents know SQL. Oracle Database Skills teach them Oracle
Getting started with our MCP Server for the Oracle Database
Introducing MCP Server for the Oracle Database
Using the Oracle SQLcl [ MCP Server|https://docs.oracle.com/en/database/oracle/sql-developer-command-line/26.1/sqcug/using-oracle-sqlcl-mcp-server.html]
LiveLabs : Exploring SQLcl [ MCP Serve|https://livelabs.oracle.com/ords/r/dbpm/livelabs/view-workshop?clear=RR,180&wid=4285] Spatial AI Thinking in theOracle AI Database:An Introduction to Spatial AI Skills](https://i.ytimg.com/vi/WrXl34n0u6A/mqdefault.jpg)





