What AI Can Tell You About Your Guns and Ammunition

Artificial Intelligence is going to take all our jobs! Well, maybe. But there’s one job I’m more than happy to turn over.

I’ll explain. My background is a little bit unusual for a “gun guy.” That’s because I’m a nerd. I spent 20 years or so in the tech business, working for geeky companies like Microsoft, some internet start-ups, and others. Even to this day, I still read tech blogs and all kinds of crazy stuff on X (formerly known as Twitter) about technology, artificial intelligence and all those dork-infested things. I find it interesting. Don’t judge me.

However, I have found a hidden benefit to this nerdy upbringing relevant to guns and shooting: record keeping.

Over the years, I’ve built all sorts of custom databases to keep track of stuff like the guns I test, the ammunition I use with them, velocities, accuracy and even penetration in ballistic gelatin. I know some folks who have said you’re not a “serious gun person” unless you track every shot you fire. I’m not of those, and I don’t really agree with that sentiment. But, given what I do for a living, it behooves me to keep gun and ammunition performance records over time, so I have.

The Value vs. Investment Tradeoff

Over the past couple of decades, I’ve tried every tech approach imaginable to solve this problem of my own making — tracking and organizing my shooting data.

Off-the-shelf software never worked for me. People who wrote those apps always seemed to have a different idea of what was important than I did.

I used multiple databases and coded my own software. I got pretty much what I wanted, but the time and effort required to make it “real software” was always more than the value of the result.

I made linked spreadsheets. Nope. Epic fail. If you’re a fellow nerd, you know why.

But now the world has changed. Artificial intelligence has breathed new life into what we can do to keep track of all this stuff. Even if you haven’t used any of these tools, you’ve likely heard about them on the news. ChatGPT. Claude Code. CODEX. Grok. CoPilot. And literally dozens more.

I won’t subject you to the geeky underbelly of how all this stuff works, just know this.

You talk to these tools like you’d talk to another human. And they understand. Not only can they answer any question imaginable, because their memories contain pretty much the entirety of the global internet, but they can also do stuff for you. Like write custom ballistics programs to track, manage and analyze whatever you want to track, manage and analyze.

Because I wanted to build a real program that not only organized all this stuff and made it easy to enter and view data, I used one called CODEX (from the ChatGPT folks) to build my personal ballistics system. CODEX is exceptionally good at, you guessed it by the name, writing computer code.

Here’s the important part. As I mentioned a second ago, you communicate with these things using everyday English. So instead of instructing it to build “SQL databases with HTML front ends,” I literally spoke it into existence. Here’s a slightly shortened version of my complete instructions.

I need to plan a new ballistics database. Don’t build anything yet until we agree on a plan and I say to proceed.

I have years of ballistics data for my writing job that is in a hacked-up spreadsheet with loosely linked tabs. It tracks guns and ammunition. Then there are three types of records I care about: Velocity, Accuracy and Penetration. The results data in those each use one gun and one ammunition type.

I need to add, change, update and delete guns, ammunition, velocity, accuracy, and penetration data.

If adding an accuracy record, I’d select a gun, and the system would show all ammunition records that match the gun’s caliber. Then I’d get a simple form to enter an accuracy record. Same for velocity and penetration records.

Now I should stop and let you ask questions and offer suggestions.

Like magic, AI understood what I wanted, and clearly understood the issues. It asked me some clarifying questions, even a few that surprised me. For example, it asked how I wanted to handle situations where a gun could fire multiple calibers. Think .38 Special and .357 Magnum, or .45 ACP and .450 SMC. Clever.

After 10-15 minutes of churning away, it declared the job done and presented my very own ballistics software, personalized for the exact wants and needs of yours truly.

Analysis By the Big Brain

Once the database exists, the value is in, you guessed it, the compiled data. Back in the old days, like months ago, I’d either code my database to calculate the simple stuff like averages or standard deviations. If I wanted to do an ad hoc study of the results, I’d export to a spreadsheet and crunch the numbers. It worked, but took some doing, and more importantly, lots of time.

My, how things have changed.

Once I had my AI buddy build and migrate all my data to a new format of its design, the real value became clear. I can retrieve whatever arcane facts strike my fancy just by talking to it. In plain English, no nerdy computer code required.

Rather than explain this, I posed a couple of random questions. Below are a few, word-for-word as I dictated them to my new best AI friend, and word-for-word how it answered.

1. Can you inspect the database and tell me what the top five most accurate handguns tested are? I’m looking for a ranking based on accuracy performance across a variety of ammunition.

Based on ammunition-balanced average MOA, the five most accurate conventional handguns are:

I normalized groups to MOA, averaged repeated groups for each gun/ammunition combination, then weighted every distinct load equally. A handgun needed at least three tested loads to qualify.

A couple of caveats:

• The Model 647 was tested at 50 and 100 yards with six-shot groups; the others largely used five-shot groups at 25 yards. MOA makes them comparable mathematically, though the protocols differ.

• The Victory has the strongest evidence: 13 loads and excellent overall performance.

• The Wilson’s fourth-place result is promising but rests on only four groups.

• If large-format pistols count, the CMMG Banshee 10mm takes first at 3.61 MOA, pushing the Staccato out of the top five.

2. Looking at .45 ACP Ammunition, do you see any measurable accuracy difference between 230-grain and 185-grain offerings?

Yes, but it’s a modest trend, not a conclusive victory.

The 185-grain loads averaged 1.40 MOA tighter — roughly 0.37″ at 25 yards.

Controlling for the firearm:

• Eight guns tested both weights.

• Six shot better with 185-grain ammunition.

• The median advantage was 1.30 MOA, or about 0.34″ at 25 yards.

• The sample is uneven, and the statistical uncertainty still includes no genuine difference.

Bottom line: this database suggests a small, fairly consistent accuracy advantage for 185-grain .45 ACP, but there isn’t enough balanced testing to declare bullet weight itself the cause. Individual load and gun preference still matter more.

3. From my recorded tests, can you tell me the average group size from the Staccato P model? Did any particular ammunition stand out in terms of accuracy?

Your Staccato 2011 P averaged 1.36 inches at 25 yards across 19 five-shot groups and seven ammunition types — equivalent to 5.19 MOA.

The best performers were:

The Black Hills Honey Badger produced both the best average and smallest group. Federal HST deserves a nod for nearly matching it over the largest sample.

No load ran away with the contest, though — the leading three were separated by only 0.12″ on average. The lone Liberty Civil Defense group, at 2.30″, was the clear laggard but needs more testing before drawing a firm conclusion.

Just to be really clear, the above results are not industry gospel, or even statistically significant in a broad sense. They are, however, accurate reflections on the particular data I happened to record at some point.

If you want to know what your guns, ammo and even your own skills have done over time, it’s a great tool. The best part? You don’t have to buy someone’s canned software and live with whatever features they thought were important. If you want a tool to track your activity over time, AI is your huckleberry.

Next up for me is a reloading module for my personalized ballistics software. I used to build that database by hand, also. Not anymore.

Get more exclusive online content from American Handgunner delivered straight to your inbox. Sign up for our free American Handgunner Insider newsletter.