AI and Automation In Action | Real-World Business Use Cases

Someone at an electronics recycling company used to spend about 9 to 10 hours every week looking up used equipment on eBay, one item at a time, just to price it. Hunter built a platform that does it in minutes, and it took him around six hours so far. In this episode, Shane and Hunter walk through how it works: you import the customer's equipment list, the platform pulls current prices and sales history from eBay, and ChatGPT analyzes the results against the company's own electronics grading guide. What we cover: - The manual process, and why it was eating a full week of someone's time - How the platform works: bid list, spreadsheet import, eBay pricing, AI analysis - Why you need to check API endpoints, not just API access - Subscription tiers, rate limits, and surprise costs - Where else this works: anything you research by hand and turn into a report - Why the best automation ideas come from the people doing the work, not leadership - How a project scoped at two months took about six hours so far CHAPTERS 0:00 Intro 0:35 The client and the challenge 1:05 The old process: pricing every item by hand on eBay 2:20 The electronics grading guide 2:45 Two API keys: eBay and ChatGPT 3:35 How the platform works 4:05 The AI analysis 4:40 9 to 10 hours a week saved 5:40 Beyond resale: any research-and-report task 6:30 The API catch: access vs. endpoints 9:00 Subscriptions, rate limits, and extra costs 9:25 Comparing prices across distributors 10:50 Who found this automation? Not leadership 12:15 Processes living in people's heads 13:40 How long did it take to build? 15:10 Scoped at two months, built in about six hours 16:15 Single sign-on with Microsoft 365 17:00 Building blocks and next steps 17:55 What should you automate first? Not sure what to automate first? Start with who does the task, how often, and how long it takes. Our Executive Jumpstart walks through exactly that. Email [email protected]. #Automation #AI #eBay

A major PHP vulnerability is all over the news, and some companies are reacting by pausing their AI work. That's the wrong takeaway. In this short briefing, Shane explains what actually happened, where AI fits in, and what to do about it today. The short version: - PHP powers roughly 70% of websites, including most WordPress sites. - The flaw is in PHP itself, not in WordPress or in AI. - It sat undiscovered for years. AI helped security researchers find it. - Updating WordPress alone won't fix it. PHP has to be patched on your server or host. What to do today: 1. Ask whoever runs your website or hosting what PHP version you're on. 2. Apply the PHP patch. 3. If you're on an old, unsupported PHP version, ask your host what your options are. CHAPTERS 0:00 Why talk about PHP on an AI podcast? 0:55 The PHP vulnerability in plain English 1:35 Why people are blaming AI 1:55 The company that paused all AI work 2:35 What the flaw actually is 3:05 It was there for years 4:00 How AI helped find it 5:15 Why this should encourage AI, not stop it 6:50 AI isn't the cause, it's helping fix it 7:25 What to do today 8:00 Wrap-up Questions about the vulnerability or how to use AI responsibly? Email [email protected]. Headline shown in the thumbnail: USA Today. #PHP #Cybersecurity #WordPress

Anyone can fake a voice or a video now. So when someone calls the help desk asking for a BitLocker recovery key, how do you know it's really them? In this episode, Shane talks with Hunter from our automation team about an automation we built for an MSP client. Before, a tech had to take the call, log into Microsoft Admin, find the device in Intune, pull the recovery key, and hand it to someone who hadn't been verified. Now the tech picks the user on a form, the user gets an approval request, and the key comes back automatically once they approve. If the user can't get to their email, the request falls over to a manager after 15 minutes. What we cover: - Why handing out a BitLocker key is a security problem, not just a time sink - How the approval step works, and how it can run through email, Teams, or SMS - 10 to 15 minutes per ticket before, under 2 minutes now - Built in under six hours on Rewst - Why you should measure a process before you automate it - Why the approval step can be reused for other automations like password resets and offboarding CHAPTERS 0:00 Intro 0:50 The problem: locked-out users and BitLocker keys 1:50 How the automation works 3:10 Why the manual process was slow and risky 4:10 Why MSPs struggle to automate (Rewst) 6:15 Deepfakes and identity verification 7:10 Before: 10 to 15 minutes per ticket 9:10 Now: under 2 minutes 10:10 Build in stages and measure first 12:15 Built in under six hours 14:10 Reusing the approval step 14:50 Wrap-up Not sure what to automate first? Our Executive Jumpstart walks through exactly that. Paying for Rewst without getting the value from it? We help with that too. Email [email protected]. #Automation #MSP #Cybersecurity

The Secure AI Playbook: Guardrails for Using ChatGPT, Claude & AI Agents at Work This episode of AI and Automation In Action presents a “secure AI playbook” for business leaders adopting tools like ChatGPT, Claude, and AI coding/agent workflows. It stresses provisioning sanctioned business accounts (not free/personal), using MFA/SSO, unique user logins, audit logs, retention and offboarding controls, and creating an AI acceptable-use policy that prohibits sharing passwords, API keys, PHI, NDA/proprietary data, and other sensitive information. The script outlines builder and agent rules such as isolating AI coding/automation on separate machines or sandboxed environments, storing credentials in vaults, using gitignore and GitHub push protection, scanning repos and packages, treating AI-generated code as untrusted, and guarding against prompt injection and risky third-party skills/connectors. It also covers least-privilege access, human-in-the-loop approvals, shadow AI scans, incident response planning, and ongoing monitoring. 00:00 Secure AI Playbook Intro 01:54 Why AI Tools Are Risky 02:50 Five-Part Framework Overview 03:46 Foundation Choose Business Plans 07:51 Account Security and Offboarding 09:59 Acceptable Use Data Policy 13:37 Builder Mode Secure Coding Setup 18:52 Secrets GitHub and Code Audits 22:41 Prompt Injection and AI Firewalls 25:11 Third-Party Skills and Connectors 27:25 Least Privilege and Human Approval 31:46 Governance Shadow AI and Incident Plan 34:49 Final Checklist and Wrap-Up

From Static Forms to a Funding Portal: Automating Nonprofit Grant Workflows with Microsoft 365 In this episode of AI and Automation In Action, Shane is joined by Hunter Holderfield to break down a real client automation for a nonprofit managing organization eligibility and funding requests. The client’s previous form tool stored submissions in SharePoint but left applicants unable to view, edit, or download what they submitted and created heavy manual back-and-forth. The team built a custom web portal with separate admin and applicant logins, Microsoft 365 single sign-on for admins, email verification for applicants, and centralized dashboards to open funding periods, review applications, approve/deny with notes, and manage organizations and users. The new solution adds form logic (including moving eligibility questions upfront), section status indicators, better communication and visibility, record retention, and the ability for applicants to propose edits for admin approval, with ongoing plans to connect further tools like SharePoint and Teams. 00:00 Show Intro and Mission 00:42 Client Story Setup 01:41 Why Holistic Automation 02:39 The Nonprofit Workflow Problem 04:15 Form Limitations and Fixes 06:17 Building the Portal Solution 10:02 Workflow Impact and Compliance 12:29 Bigger Lessons and Fast Delivery 14:16 Future Enhancements and SharePoint 15:36 Wrap Up and Call to Action