When I was data consulting, the companies we worked with didn’t have technical bottlenecks. They had folks writing SQL and leveraging FiveTran or DBT or implementing Snowflake. Their issue was understanding what the data was, what it was supposed to be, and what the my needed it to do. Those are people problems.
X is usually “killed” because its would-be killers see only an application-centric world, despite the demonstrable benefits of infrastructure built for loose coupling using open standards.
SQL isn’t going anywhere because, globally, systems of record run on RDBMS platforms that are extremely costly to replace. Today, many organizations still run ancient RDBMS releases, at least from the vendors’ perspective, because customers are reluctant to upgrade.
Is there a problem with SQL and the relational tables it operates on in the age of AI? Of course. But ripping and replacing an RDBMS isn’t a practical solution.
This is exactly where open-standards-based loose coupling helps, courtesy of ODBC and JDBC as database connectivity standards for loosely coupling AI agents, Skills, and MCP-accessible tools with DBMS platforms such as Oracle, DB2, SQL Server, Sybase, Ingres, Informix, Progress/OpenEdge, and others. You can even bridge ODBC to JDBC and vice versa.
The point is that once connected, you have a foundation for building Context Layers informed by Ontologies that enable more deterministic outcomes when working with AI agents, Skills, and Tools.
Joe, I will be shamelessly quoting this article in my company meetings and here on the stack. Fantastic article, absolutely loved the marketing angle and that hype is just hype!
I thought you were talking about X (aka Twitter) here for a sec. But read the article, and definitely agree, the this is dead take is way overdone and never really true…
When I was data consulting, the companies we worked with didn’t have technical bottlenecks. They had folks writing SQL and leveraging FiveTran or DBT or implementing Snowflake. Their issue was understanding what the data was, what it was supposed to be, and what the my needed it to do. Those are people problems.
Bingo
AI is dead. Just like data modeling and SQL
X is usually “killed” because its would-be killers see only an application-centric world, despite the demonstrable benefits of infrastructure built for loose coupling using open standards.
SQL isn’t going anywhere because, globally, systems of record run on RDBMS platforms that are extremely costly to replace. Today, many organizations still run ancient RDBMS releases, at least from the vendors’ perspective, because customers are reluctant to upgrade.
Is there a problem with SQL and the relational tables it operates on in the age of AI? Of course. But ripping and replacing an RDBMS isn’t a practical solution.
This is exactly where open-standards-based loose coupling helps, courtesy of ODBC and JDBC as database connectivity standards for loosely coupling AI agents, Skills, and MCP-accessible tools with DBMS platforms such as Oracle, DB2, SQL Server, Sybase, Ingres, Informix, Progress/OpenEdge, and others. You can even bridge ODBC to JDBC and vice versa.
The point is that once connected, you have a foundation for building Context Layers informed by Ontologies that enable more deterministic outcomes when working with AI agents, Skills, and Tools.
SeeAlso:
[1] What an AI Agent can do with a CSV file -- https://www.openlinksw.com/weblog/?post=2026-practical-data-community-state-of-data-engineering-gemini.html
[2] Simple SQL Server showcase re Knowledge Graphs informed by an Ontology generated on-the-fly from ODBC-accessible tables -- https://www.openlinksw.com/weblog/?post=rdf-linked-data-views-sqlserver.html
Kimball and Ross.
I hope to see the day when "X" = "Hype", but at the moment it seems to be a very unlikely event.
If I may introduce an article I already wrote more than a year ago on these patterns - Things not to do as a Data Influencer on LinkedIn: https://peterbaumann.substack.com/p/things-not-to-do-as-a-data-influencer
Snap! https://greyskullanalytics.substack.com/p/articles-that-proclaim-something?utm_source=share&utm_medium=android&r=2vhsqj
Joe, I will be shamelessly quoting this article in my company meetings and here on the stack. Fantastic article, absolutely loved the marketing angle and that hype is just hype!
hells yeah
I thought you were talking about X (aka Twitter) here for a sec. But read the article, and definitely agree, the this is dead take is way overdone and never really true…
ha, yeah, people can defintely interpret it as the social network. Which is sorta true in it's own way.