Justin Dews, PathOpt | Responsible AI & How To Use It

Meet Justin Dews

For our first edition of Comms Center Stories, we sat down with Mr. Justin Dews – the CEO and co-founder of PathOpt, a firm specializing in agentic AI solutions for workflow automation and onboarding for small businesses.

Based in Dallas-Fort Worth, Dews has a 20+ year background in Information Technology, working with clients in trades and construction until pivoting in 2024 to launch PathOpt.

Now in 2026, PathOpt innovates by developing long-term solutions to preserve “tribal knowledge” – that is, the most crucial on-the-job knowledge that is seldom written down – saving small businesses time, effort, and revenue traditionally lost in onboarding.

First perceptions of AI models

In 2022, the first public release of OpenAI’s ChatGPT research model marked the beginning of the LLM “space race” we find ourselves in now. In retrospect, those initial models were substantially less-sophisticated than what we’ve seen now with the likes of Fable and Mythos.

Asking Dews about his initial impressions of AI back then, he gave a similar outlook: the early models for ChatGPT and Cursor were promising, but had yet to demonstrate their effectiveness for much beyond novel work/experimentation.

Dews’ perspective on AI changed in 2025 after the most recent developments in agentic AI – that is, AI that can be trained to autonomously handle specific tasks with persistent planning and memory.

Useful AI vs. timesink

For Dews, he’s betting on the power of AI agents as invaluable assistants that simplify workflows for good, not as replacements for human input:

What AI is really great at … is drawing from massive stores of data to answer specific questions.

Instead of treating AI tools as independently intelligent, Dews focuses PathOpt on the analysis of preexisting and current data to build useful models.

He further reminds us that for an AI agent to be useful to a business, it has to be properly trained by a human with the experience to define a quality job in the first place – then iterate.

Advancing tried and true work into an accessible age

As we discussed the similarities between PathOpt and Comms Center, we found both of our missions to point towards a goal of streamlining chaos and tedium that “comes with the territory” in these industries.

In Dews’ case, he views AI as a revolutionary force across all industries – enabling business to use modulated agents appropriate to their scale to tackle the most tedious and presence-intensive tasks.

Particularly with respect to onboarding, Dews points to just how difficult it is to train the new hire: you not only need to have SOPs ready, but you’ll need the person who came up with them in the first place to be free enough to teach them.

Here, PathOpt works directly with clients to source and gather the data needed for a task like onboarding, billing, or estimating to tailor a secure AI agent that can interface internally or to the client.

Once finished, the agent then acts as a “clone” of the employee with the necessary experience: allowing new hires or clients to use the AI model to answer crucial questions, all while returning time and effort back to the business.

How do we prepare for an AI future?

Dews reassures us that though AI brings new vulnerabilities and technical challenges, we may still rely on the same fundamentals of security that we employ at our businesses every day.

Simply put: don’t give an AI agent any more privileges/access than you would for a recent employee at your company. Use strong passwords and rotate them, enable two-factor authentication, and please don’t post your keys publicly.

At the end of the day, the trades and construction projects all require a human touch to account for mistakes, sudden changes, and creativity. AI won’t account for these demands, but it surely can help us focus on these important items instead of the tedium of an automatable task.