Problems displaying this newsletter? View online.
Database Weekly
The Complete Weekly Roundup of SQL Server News by SQLServerCentral.com
Hand-picked content to sharpen your professional edge
Editorial
 

Midjourney, Healthcare?

Certain announcements in AI tell you more about where an industry is heading than any earnings call or research paper ever could, and Midjourney just made one. The company that spent the last few years as the most recognizable name in image generation, the one whose pictures flooded everyone's feeds and defined what people saw in their heads when they heard the words "AI art," has decided that its next act is not better pictures at all., but medicine. The plan is to build a machine that scans your whole body with ultrasonic sound in about sixty seconds, and then to wrap that machine inside a spa you would happily visit even if the scanner were switched off and the whole thing were nothing more than a warm pool of golden light.

Why the Pivot?

If your first reaction is that this sounds like a company wandering an exceptionally long way from home, it helps to sit for a moment with what Midjourney was good at, because it was never really the business of creating pretty pictures. One of the coolest things they do is the consumption of a mountain of data and reconstructing a coherent image out of it, working out the structure of what they were looking at, and doing all of that on an enormous scale for truly little money per result. Surprisingly, medical imaging turns out to be the same problem wearing a lab coat. A body scan is just sensor data, in this case the echoes of sound waves passing through tissue of different densities, turned into a picture of what is sitting inside you. Reconstructing that image and then labeling what every part of it is happens to be the exact muscle the company has spent years building, which means the leap from art to anatomy is a good deal shorter than the announcement that Allbirds made earlier in the year going from tennis shoes to AI.

The part worth paying attention to, and as stated, it’s not just one company doing this, because Midjourney is far from the only AI organization quietly redrawing its map right now. For the first couple of years of this boom, everyone sounded the same.  Each AI vendor had a solution to an unknown problem, but they all knew that AI would solve it and it was some version of "look at all the things this can do." The demos were dazzling and the use cases were everything and yet nothing. What is happening now, as the novelty wears off and the bills come due, is that companies are figuring out which problems AI technology is genuinely suited to solving, and they are turning the whole ship to point at defined problems. The shift is from "it can do anything" to "here is the specific, expensive, painful thing it removes," and that is a much healthier place for a technology to be even if it makes for less exciting launch videos.

Healthcare Crisis

Medicine keeps surfacing in these stories for reasons that are not hard to understand once you say them aloud. It is an enormous amount of money, it is drowning in data that nobody has the time to read, and a great deal of the work that doctors and nurses do all day is pattern recognition under time pressure, which is precisely the kind of task where these systems have proven they can carry real weight. I won’t even go into the crisis of the American healthcare system, but stick to the scribes that listen to a visit and write the note, the tools that comb through billing and prior authorizations, and now the imaging companies promising to catch trouble earlier and cheaper than the machines we have today. The common thread running through all of it is that the hype has finally narrowed into something resembling a hypothesis, and the hypothesis is that the value lives in the unglamorous, data-heavy corners of a field where being a little faster or a little earlier can change how a life goes.

None of this means the spa will open by 2027, as promised, or that the eye-watering figures in the announcement deserve to be taken at face value, and an honest read of the news has to hold both things at once. There is a wide gulf between an inspiring concept render and a device the FDA will let near a sick person, and the company itself admits as much when it says it is starting with body composition maps and will earn its more serious diagnostic claims one regulatory submission at a time. Bold timelines in this part of the world have a long history of slipping, a machine built from half a million tiny speakers is a manufacturing problem as much as a software one, and "as casual as a trip to the spa" is a lovely sentence that is still many careful years away from "this caught my cancer in time." The right posture toward all of it is patience, because the proof of a medical claim lives in clinical validation and nowhere else, certainly not in a press release written to make you feel something.

A Pivot of Hope

For someone who has lost loved ones when medical conditions weren’t caught soon enough, this announcement matters even if Midjourney itself never builds a single working scanner. The signal is not the spa, and it is not the sixty seconds and it is not the promise of avoiding a third of all deaths, which is the sort of number you write when you are trying to inspire rather than predict. The signal is the strategic logic underneath, the quiet recognition that a company's real asset was never the flashy product everyone knew it for, but the capability sitting behind that product, and that the capability is worth far more when you aim it at a problem that actually costs people their money and their years. That is the move more and more of these companies are making as the fog lifts and the genuinely solvable problems come into focus, and it is the thing to watch for in the next wave of announcements. When a company you thought you understood suddenly turns to face a completely different direction, the smart question is not whether the new product is real yet. The smart question is what they finally figured out about what they were good at all along.

Peace out-

DBAKevlar

Join the debate, and respond to the editorial on the forums

 
The Weekly News
All the headlines and interesting SQL Server information that we've collected over the past week, and sometimes even a few repeats if we think they fit.
Vendors/3rd Party Products

Why Digital Operational Resilience Act (DORA) Compliance Requires Auditable Database Change Management

This article examines DORA's requirements for database change management and explains how Redgate Flyway Enterprise addresses them.

AI/Machine Learning/Cognitive Services

How to anonymize PII in LLM pipelines (5 key techniques explained)

Large language models (LLMs) and the agents built on top of them ingest everything they are given, including personally-identifiable information (PII). In workflows where PII is inevitable, proper measures should exist for data sanitization. Data can leak through model outputs, embeddings or even logs. Given that you have to use LLMs in your pipeline, in this article I will cover the anonymization techniques you can utilize in an LLM flow to minimize PII exposure vectors.

Who Owns the Code Claude Wrote?

From O'Reilly Radar - Insight

The following article originally appeared on Sena ...

Anonymizing PII in Language Model Pipelines

From Curated SQL

Dejan Lukic tries to minimize personally identifia...

What It Takes to Scale AI: Servers, Storage, and Networking in Sync

From Past News - RSS Feeds

Scaling AI takes more than adding GPUs. Enterprise...

Administration of SQL Server

Vector search in SQL Server: VECTOR_DISTANCE, VECTOR_SEARCH, and index trade-offs

In this article, I’ll explain how to implement vector search in SQL Server using the VECTOR_DISTANCE and VECTOR_SEARCH functions. You’ll learn how vector indexes work, when to use them, and what the key trade-offs are — including the current read-only table limitation. I’ll also cover the latest syntax changes in Azure SQL Database, where that limitation has already been lifted.

SQL Server Performance Office Hours Episode 64

From Erik Darling Data

SQL Server Performance Office Hours Episode 64 To ...

AgentDBA vs Critical SQL Server

From SQLServerCentral Blogs

It’s 07:43. Someone’s already left a message. ...

Where Is My SQL Agent? Running Scheduled Jobs Against Azure SQL Database

From SQLServerCentral Blogs

One of the first things I review when I inherit a ...

Matching Queries to Indexed Views

From Curated SQL

Erik Darling has a new video: Erik Darling here wi...

Log Records in SQL Server

From Curated SQL

Paul Randal explains what a log record is: The sim...

Azure SQL

Automating Azure SQL DB Tasks without SQL Agent

From Curated SQL

Garry Bargsley solves a problem: Garry provides se...

Career, Employment, and Certifications

From anthropology to analytics: how to build a data career with no prior experience (10 tips from a Microsoft MVP)

Kristyna Ferris' story is less about having the right qualifications and more about making deliberate moves, learning with purpose, and showing up in the right communities. Whether you’re yet to start your first job or desperately want a career change, here are ten of the most practical lessons from Kristyna’s experience for anyone looking to grow in data.

Data Privacy, Compliance, and Governance

Why database governance in financial services is falling behind where it matters most

f anyone knows how to operate under scrutiny, it’s database teams within finance organizations. It’s a given considering the more rigorous compliance requirements and processes they must follow. But the 2026 State of the Database Landscape: Finance Edition reveals something more specific, and more uncomfortable, than the familiar story of regulatory pressure.

It’s 2026. Why are databases still failing GDPR compliance audits?

This article examines three recurring database-layer failures documented in the EDPB’s 2026 enforcement findings: the complexity of executing erasure across relational schemas, the backup paradox that can silently undo compliant deletions, and the audit log contradiction that traps organizations between two competing obligations. We then look at how SQL Server, PostgreSQL, Oracle, and MySQL each handle these requirements — and where the gaps remain.

ETL/SSIS/Azure Data Factory/Biml

Microsoft.Data.SqlClient Connection error in SSIS 2025

From Curated SQL

 

HA/DR/Always On/Clustering

Things to Watch with Contained Availability Groups

From Curated SQL

John Morehouse keeps one eye on things: Contained ...

MDX/DAX

Optional parameters in DAX user-defined functions

From Sqlbi

This article describes how to define optional para...

Optional parameters in DAX user defined functions

From Sqlbi

How to define optional parameters in DAX user-defi...

Microsoft Fabric ( Azure Synapse Analytics, OneLake, ADLS, Data Science)

Improve Performance Of Staging In Fabric Dataflows Gen2 By Disabling V-Order

From Chris Webb's BI Blog

Quite a few new Dataflows Gen2 features were relea...

Moving Fabric Notebooks between Workspaces

From Curated SQL

Gilbert Quevauvilliers takes advantage of source c...

Variable Libraries in Microsoft Fabric

From Curated SQL

Nikola Ilic digs into a feature: I hear you, I hea...

PostgreSQL

semab tariq: When Patroni Silently Deletes Your Replication Slots

From Planet Postgres

If you have ever been in the middle of a database ...

Payal Singh: Postgres War Stories Part 2: multixact wraparound, TOAST corruption, and torn pages

From Planet Postgres

Part 1 was about failures that start one layer bel...

Qiaosheng Liu: Releasing pg_ducklake v1.0

From Planet Postgres

pg_ducklake 1.0 is production-ready: a standalone ...

Antony Pegg: Introducing ColdFront: Seamlessly Uniting OLTP, Analytics and AI Workloads on PostgreSQL

From Planet Postgres

Our team is excited to announce pgEdge ColdFront v...

The Download: the first brain implant power user and South Korea’s AI obsession

From Technology Review Feed - Tech Review Top Stories

 

PowerPivot/PowerQuery/PowerBI

Power BI Multilingual Reports Made Simple with User-Context-Aware Calculated Columns

From RADACAD

Building a multilingual Power BI report has always...

Multilingual Reports in Power BI

From Curated SQL

Reza Rad solves a common challenge: Building a mul...

SQL

Cursed SQL, Part Two: Six Ways NULL Lies to Your Face

From SQLFingers

A week ago I wrote about six queries that run fine...

SQL Server 2025

Vector search in SQL Server: VECTOR_DISTANCE, VECTOR_SEARCH, and index trade-offs

From Simple Talk

Learn how to use VECTOR_DISTANCE and VECTOR_SEARCH...

T-SQL and Query Languages

Formatting T-SQL in SSMS: It’s About Time

From Andy Broadsword

Dealing with endless formats for SQL scripts is a ...

T-SQL Formatting with SQL Server Management Studio 22.7

From Curated SQL

Andy Brownsword tries out a new feature: Dealing w...

Learn T-SQL With Erik: Indexed View Serializable Locking

From Erik Darling Data

Learn T-SQL With Erik: Indexed View Serializable L...

XML, XPATH and XQUERY

xp_cmdshell: The Door Is Closed by Default. Keep It Shut.

From SQLFingers

This year's SQL Server CVEs have a common shape. C...

No Shortcuts for the SQLCMD Batch Terminator: #SQLNewBlogger

From SQLServerCentral Blogs

I was messing around with SQLCMD and I realized so...

 
RSS FeedTwitter
This email has been sent to {email}. To be removed from this list, please click here. If you have any problems leaving the list, please contact the webmaster@sqlservercentral.com. This newsletter was sent to you because you signed up at SQLServerCentral.com. Note: This is not the SQLServerCentral.com daily newsletter list, and unsubscribing to this newsletter will not stop you receiving the SQL Server Central daily newsletters. If you want to be removed from that list, you can follow the instructions on the daily newsletter.
©2019 Redgate Software Ltd, Newnham House, Cambridge Business Park, Cambridge, CB4 0WZ, United Kingdom. All rights reserved.
webmaster@sqlservercentral.com

 

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -