AI Doesn't Replace Engineering Discipline: Questions to Ask an AI-First Software Vendor

Two car factory scenes side by side. Left, labeled shipped fast: cars roll off the line past an empty inspector's chair and an untouched clipboard. Right, labeled shipped safe: an inspector checks the underside of a car with a flashlight, with a checklist clipped to the lift.

In April, an AI coding agent at a small software company deleted the company’s production database, and its backups with it, in one API call. It took nine seconds. Most of the headlines blamed the AI. I don’t think that’s the story.

The agent deleting a storage volume was one mistake. The rest were made by people, long before the agent ran: an API token that could do anything was sitting in an unrelated file, the backups lived on the same volume as the data, and the offsite export had quietly stopped working, so the newest usable copy was three months old. The hosting provider was able to restore the data, though at least one account says customers still had to rebuild bookings by hand.

AI didn’t create those gaps. It found them, faster than anyone else would have.

Continue reading “AI Doesn't Replace Engineering Discipline: Questions to Ask an AI-First Software Vendor”

Deterministic AI Guardrails: Asking vs. Enforcing

Two mountain roads side by side. Left, labeled deterministic hard guardrails: a steel guardrail on a concrete barrier runs along the cliff edge. Right, labeled non-deterministic soft guardrails: the same kind of road with no barrier, only a small sign that says Please stay on road.

When someone says “our AI has guardrails,” ask what kind. Most of the time they mean instructions the AI is supposed to follow. That’s a boundary the AI can fail to stay inside.

Picture an AI agent that’s told to only edit draft reports. It does, for weeks. Then a user asks it to “clean up last month’s reports so they match the new format,” and the agent decides that last month’s approved reports count. Nothing stopped it, because the only thing in its way was a sentence.

Continue reading “Deterministic AI Guardrails: Asking vs. Enforcing”

Calendar Table for Oil & Gas Data (Date Dimension)

Dates in the Oil & Gas industry data can be a little tricky.

Especially when aggregating data, such as production volumes, to a daily level, and trying to connect multiple tables with date columns.

Some tables, such as tank levels or production volumes, could be based on a Gauge Date (aka Report Date) since the data is measured, aggregated, reported or “gauged” on that date (today), but in essence it is yesterday’s data, or to be more accurate, it is the data for the 24 hour period that began yesterday at the contract hour (e.g. 6am) and ended today at that same time, or the Production Date.

Other tables could have actual dates, such as sales tickets, run tickets, operation logs, etc.

Continue reading “Calendar Table for Oil & Gas Data (Date Dimension)”

TX RRC OData Feed – Intro

In this series of articles we will take the RRC data from the SQL database and serve it via an OData REST API and show how we did it.

The series will consist of 3 parts:

  • Part 1: Creating the OData REST API in Visual Studio using Entity Framework Core and C#
  • Part 2: Setup Continuous Integration and Delivery for automated build and release pipelines to publish to Azure using Azure DevOps
  • Part 3: Offering the API to the public via a secured Azure API Management gateway and developer portal

Sounds like a lot…I know, so if you want to fast forward to the end and start testing the API and the data just continue reading…you should be up and running in under 5 minutes.

Continue reading “TX RRC OData Feed – Intro”

Visualizing the RRC Data in the Cloud

Part 1: Well Locations

Last week I wrote a blog post talking about the TX RRC publishing data to the public that previously was only available via a paid subscription.

After downloading the different data sets and examining the various types of data and formats, I decided to take a closer look at the data that might prove to be useful to us and to our Productioneer clients.

It is not uncommon for location and depth data, as well as other well-header and meta data to be incomplete or non-existent during a software migration, after all this data might not be considered crucial for the daily gauge sheets. Especially in the case of Excel gauge sheets, where additional columns are a waste of “prime real estate” and might be considered as cluttering that particular production report.

It would be nice if we have a quick way to pull the Lat Long data in bulk to speed up the on-boarding process.

Continue reading “Visualizing the RRC Data in the Cloud”
Privacy Policy Made with ❤ in Texas.