Is Your AI Actually Helping You — Or Just Sounding Helpful?
There's a difference between an AI that gives you an answer and an AI that gives you the right answer. Most business owners have never been shown that difference. They type a question, get a response, and move on — never realizing that a small shift in how they communicate with AI could multiply the quality of every output they receive, save hours of rework every week, and in some cases, protect them from acting on information that was simply, confidently, completely wrong.
That's exactly what Episode 89 of Unhacked is about.
The AI Prompting Frameworks That Change Everything
Josh Holloway of Seventh Eye Technologies introduces the concept of structured AI prompting using proven frameworks — and explains why most people are treating AI like a search engine when they should be treating it like a highly capable, context-hungry collaborator.
The central framework covered is PTCF: Persona, Task, Context, and Format.
- Persona — Tell the AI who it is. "You are an expert cybersecurity professional with a CISSP certification and 25 years of experience." This single step transforms generic output into expert-level guidance.
- Task — Be specific about what you want done. Not "tell me about incident response" but "build an incident response playbook for business email compromise, to be used by IT and management staff during a high-anxiety crisis event."
- Context — Give it background: your industry, your company size, your audience, your constraints, your existing policies. The more context it has, the more precisely it can serve you.
- Format — Tell it how you want the output delivered. A Word document. An outline. A summary. A step-by-step checklist. Without this, AI decides for you — and it won't always decide well.
The result of applying this framework, demonstrated live on the episode: a generic prompt produced a 12-page document. The same request, reframed with PTCF, produced a 24-page, professionally structured incident response playbook — complete with cover page, table of contents, acceptance policies, ownership assignments, internal response procedures, and appendices. Same AI. Same topic. Radically different outcome.
Other Frameworks Worth Knowing
Josh also walks through several additional frameworks your team can start using immediately:
- TAG — Task, Action, Goal
- BAB — Before, After, Bridge
- CARE — Contact, Context, Action, Result, Expectation
- GROW — Goal, Reason, Output, and more
These frameworks will be available as a downloadable resource in the Unhacked portal.
The "Use AI to Use AI" Technique
Brian Lachapelle of B4 Networks shares a technique that sounds almost too simple — but produces dramatically better results: if you don't know how to structure your prompt, ask the AI to write the prompt for you. Describe what you're trying to accomplish and say: "Can you help me draft the best prompt to accomplish this?" Then feed that generated prompt back in. The difference in output quality is consistently significant, and it's a technique anyone can use on day one with zero AI experience.
The Custom GPT / Project Shortcut
One of the most practical takeaways from this episode: you don't have to re-explain yourself to AI every single time you use it. Both ChatGPT (Custom GPTs) and Claude (Projects) allow you to build a persistent context — a set of instructions, a defined persona, a set of uploaded documents — that the AI remembers every time you open it.
Brian demonstrates this with a real example: he uploaded his entire company employee manual and all policies to a Custom GPT, then instructed it to answer employee questions based solely on those documents. Now instead of staff digging through binders or emailing HR for basic policy questions, they get instant, accurate answers — locked down to company-approved content only, with no internet access.
Justin adds his own lesson learned the hard way: he had been using AI to generate episode outlines for dozens of episodes, then manually reformatting everything to match his standard template — a task that was costing him 15–20 minutes every single week. Once he uploaded his template and told the AI to match it, that work disappeared entirely. The fix took seconds. The wasted time had been accumulating for months.
Hatz: Secure AI for Business Teams
The episode also features a walkthrough of Hatz, a business-focused AI platform that solves one of the most critical problems with consumer AI tools: data security.
When your employees use ChatGPT or Claude through a personal account, you have no control over what they're submitting — client names, financial data, contracts, sensitive communications could all be leaving your organization with zero visibility. Hatz changes that. It provides:
- A contained environment where sensitive data doesn't leak into AI training systems
- The ability to manage which employees have access and to what
- A data filter that prevents sensitive information (like Social Security numbers) from being transmitted
- Access to virtually all major LLMs (Claude, GPT, and many more) from one platform
- A workflow builder that allows multi-step AI processes with different models at each step
- An "auto" mode that intelligently selects the best AI model for each specific task
- Webhook integrations that let other systems trigger AI workflows automatically
For business owners who want to embrace AI without gambling with client data or company secrets, Hatz represents the kind of secure-by-design approach that responsible adoption requires.
AI Hallucinations Are a Real Business Risk
Perhaps the most important — and most underappreciated — point in this episode is this: AI will be wrong. Confidently, fluently, persuasively wrong.
The hosts share real examples from their own lives. Mario asked ChatGPT for the exact recipe for a fondue dish from his wife's favorite restaurant. He bought every ingredient on the list. His wife immediately recognized that the recipe was wrong. When Mario pointed out the errors, the AI kept agreeing — "Great catch!" — which revealed that it had never actually known the correct recipe to begin with. It had fabricated plausible-sounding answers and then endorsed his corrections as if it had known the truth all along.
Brian had the same experience: ChatGPT included ingredients in his shopping list that it later told him had nothing to do with the recipe. Justin's AI confidently told him Nevada is in the Mountain Time Zone. (It isn't.)
These are cooking and geography anecdotes — low stakes. But AI is being used today to review contracts, draft policies, analyze compliance requirements, and write code. In those contexts, a hallucinated clause, a fabricated regulation, or a flawed code block can translate directly into legal exposure, security vulnerabilities, or costly operational failures.
The lesson: AI is a tool, not an authority. Verify. Always.
The hosts of Unhacked are not AI enthusiasts who are excited about technology for its own sake. They are business owners and IT professionals who have spent decades protecting the companies of people just like you. Their enthusiasm for AI is grounded in one specific reality: when used correctly, AI gives small and mid-sized businesses access to capabilities that were previously only available to enterprises with large teams and large budgets.
But "when used correctly" is doing a lot of work in that sentence.
Used carelessly, AI creates new risks: data leakage, over-reliance on unverified outputs, security gaps from unsanctioned tools in employee hands, and the false confidence that comes from a tool that sounds authoritative even when it's wrong.
The frameworks in this episode are not just productivity tips. They are the foundation of responsible AI adoption — the difference between a tool that works for your business and one that quietly works against it.
🔒 Is Your Business Actually Protected?
The conversations on Unhacked exist for one reason: to make sure that business owners like you don't become the next cautionary tale. One phishing email, one unsecured AI integration, one data breach — and everything you've built is at risk.
The single most important step you can take today is to find out where your vulnerabilities actually are.
Schedule your FREE Security Assessment with Phoenix IT Advisors →
Our team will walk through your current environment, identify your real exposure, and give you a clear picture of what it would take to protect what you've built. No pressure. No jargon. Just clarity.
Because the cost of finding out is zero. The cost of finding out too late is everything.
Schedule Your Free Security Assessment - and mention UnHacked.