OnPath Testing

OnPath Testing OnPath Testing offers software testing QA services for companies developing desktop, web, and mobile applications.

OnPath Testing helps you create a successful product launch by offering the following software testing services:

* Review your testing needs
* Write a test plan approved by you
* Perform manual and automated tests
* Report issues to your team
* Assist you with your project roll-out and deployment

Our trained testing professionals are active team members and capable of participating in all

product phases. We are experienced in the latest Quality Assurance (QA) methodologies, and integrate smoothly with your development team. We apply our experience and management of offshore resources to offer you extensive technical expertise at the best price.

08/31/2026

AI-generated code now accounts for over 40% of what's getting committed to production. That number has been climbing fast.

Here's what's not climbing: developer confidence in it. Trust in AI code accuracy sits around 29% right now, down from 40% not long ago. Meanwhile, 66% of developers say their biggest frustration is code that's "almost right but not quite."

Almost right is the most expensive kind of wrong in software. It passes a quick look. It clears a rushed review. It gets merged. Then it fails at 2am in a way nobody anticipated.

After almost 30 years in dev and QA, I can tell you that "almost right" has always been harder to catch than flat-out broken. Broken is obvious. Almost right requires someone who's actually looking.

That's what QA is for. Not to slow things down. To make the "almost" visible before it ships.

08/30/2026

Something I've been thinking about this weekend:

We talk a lot about what AI can build. We don't talk nearly enough about what happens when what it built is wrong and nobody was watching.

Not catastrophically wrong. Quietly wrong. The kind of wrong that shows up three releases later and takes a week to trace back to its origin.

Almost 30 years building and testing software, and the bugs that have cost teams the most were never the dramatic ones. They were the ones that slipped through because everyone assumed someone else had checked.

Have a good rest of your weekend. Go outside. The code will still be there.

08/29/2026

It's Saturday, so here's a completely serious and professional poll conducted by me, just now, in my head:

What do you do when an AI generates code that technically works but is so bizarre you can't explain it to anyone?

A) Ship it, it works
B) Rewrite it so the PR review doesn't raise questions
C) Leave a comment that says "don't touch this"
D) Stare at it for 20 minutes and then walk away

There is no wrong answer. There are only answers that are more honest than others.

08/28/2026

QA roles grew 17% over the past couple of years. Developer roles grew 9%.

That gap is not a coincidence. It tracks almost exactly with how fast AI coding tools went mainstream.

The volume of code being written has gone up. The complexity of what's being shipped has gone up. The number of people whose job it is to verify that code actually works has, correctly, gone up too.

For teams that have been operating without dedicated QA because they thought they couldn't afford it, or because they assumed the developers would catch everything, that gap is getting harder to justify.

Almost 20 years running OnPath, and I've never once seen a team that said they wished they'd started QA later.

08/26/2026

A study of 5,600 vibe-coded production apps found 2,038 critical vulnerabilities, over 400 exposed secrets, and 175 PII exposures.

Those aren't prototype apps. Those are live, user-facing products.

The pattern is pretty consistent: a small team uses AI to move fast, ships something that works, and skips the part where someone checks whether it's also secure. Not because they're careless. Because they're stretched, the feature worked in testing, and there was never a dedicated moment to ask "what are we missing."

That dedicated moment is exactly what a QA function is supposed to provide. It doesn't have to be a large team or a long process. It just has to be intentional.

Speed is the point of AI-assisted development. But speed without a checkpoint isn't velocity. It's drift.

08/25/2026

Ranking the stages of grief when the AI writes your code:

1. Awe (it did it so fast)
2. Confidence (this is clearly correct)
3. Mild concern (what is this function actually doing)
4. Bargaining (maybe it's fine)
5. A debugging session that takes longer than writing it yourself would have
6. Acceptance
7. Writing a test you probably should have written first

We're all somewhere on this list. Some of us just live at step 5.

Adding more tests won’t fix broken quality.If your QA process feels slow, reactive, or unreliable… the problem isn’t eff...
05/11/2026

Adding more tests won’t fix broken quality.

If your QA process feels slow, reactive, or unreliable… the problem isn’t effort. It’s structure.

The Test Pyramid helps you:

- Catch issues earlier
- Reduce reliance on slow UI tests
- Build confidence before release

Read the full blog to learn how to implement it correctly and why this test is still relevant today: https://www.onpathtesting.com/blog/the-test-pyramid-in-2026-still-relevant-still-necessary/

05/11/2026

After almost 30 years in dev and QA, I've learned to spot the warning signs.

When your product manager says "just one tiny feature" on Friday at 4 PM.

When the deadline is "whenever you're done" but also "ASAP."

When stakeholders ask if we can "just skip testing this time" because it's a "simple change."

(Spoiler alert: there's no such thing as a simple change.)

What are your favorite famous last words in software development?

05/10/2026

Sunday reflection: Nearly 30 years building software, and I'm still amazed by how much trust we place in code.

Every app launch, every online purchase, every message sent — we're trusting thousands of lines of logic written by humans like us. Humans who have bad days, make typos, and sometimes forget edge cases.

Yet somehow, most of the time, it all just works.

That's not magic. That's the result of countless QA professionals, code reviewers, and developers who care deeply about getting it right.

To everyone who's ever caught a bug before it reached production: you're the reason the digital world keeps spinning.

Here's to another week of building software that works.

05/09/2026

Saturday morning thought: Why do we call them "bugs" anyway?

Grace Hopper found that famous moth in a Harvard Mark II computer back in 1947, but programmers were calling problems "bugs" even before that.

Maybe because bugs are everywhere, hard to predict, and show up when you least expect them?

Or because fixing one sometimes reveals three more hiding nearby?

Either way, I'm grateful for the metaphor. "Debug" sounds so much better than "de-unexpected-software-behavior."

What's the weirdest bug you've encountered lately?

Address

Boulder, CO

Telephone

(303) 479-4994

Website

Alerts

Be the first to know and let us send you an email when OnPath Testing posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Contact The Business

Send a message to OnPath Testing:

Shortcuts

Share

Category