{
 "byKey": {
  "cover": [
   {
    "say": "Welcome back to Practical AI for Business. This is our final chapter, and it's about rolling AI out safely across your team.",
    "item": 0
   },
   {
    "say": "We'll cover three security layers, a three-step adoption plan, and then a recap of the whole course.",
    "item": 0
   }
  ],
  "EP11 | Three Security Layers for Using AI at Work#1": [
   {
    "say": "Let's start with episode eleven, and how you'd shoot its sixty-second video.",
    "item": 0
   },
   {
    "say": "The hook asks: pasting financial data straight into a free AI tool? Your company secrets could end up where they shouldn't.",
    "item": 1
   },
   {
    "say": "The pain point is a familiar one. To hit a deadline, people use personal AI accounts for customer lists and financial reports, and the company has no idea where that data goes.",
    "item": 2
   },
   {
    "say": "In the hands-on part, you show three things: a company-approved account, the data-use settings, and masking sensitive data before you paste.",
    "item": 3
   }
  ],
  "EP11 | Three Security Layers for Using AI at Work#2": [
   {
    "say": "The video wraps up with a simple message.",
    "item": 0
   },
   {
    "say": "The answer is not to ban AI. It's to set rules for using it safely, and then invite viewers to comment for an AI use policy template.",
    "item": 1
   }
  ],
  "EP11 | Three Security Layers for Using AI at Work#3": [
   {
    "say": "Now, a quick correction to that hook.",
    "item": 0
   },
   {
    "say": "Saying company secrets become public training material is dramatic, but it's not the full picture. Data-use policies differ by service and by plan, and they change over time.",
    "item": 1
   },
   {
    "say": "Some services don't train on your content by default, some give you an opt-out, and some business plans include contractual guarantees.",
    "item": 1
   },
   {
    "say": "So don't trust everything, and don't panic about everything. Check the current terms and settings, and treat what's in writing as the source of truth.",
    "item": 1
   }
  ],
  "EP11 | Three Security Layers for Using AI at Work#4": [
   {
    "say": "Who is this for?",
    "item": 0
   },
   {
    "say": "Any company where people already use AI, or plan to, at any size.",
    "item": 1
   },
   {
    "say": "It matters most if you handle customer personal data, finance, research and development, or contract information.",
    "item": 1
   }
  ],
  "EP11 | Three Security Layers for Using AI at Work#5": [
   {
    "say": "Here are the three security layers. Layer one is data classification and the minimum-data principle.",
    "item": 0
   },
   {
    "say": "First, sort your data into levels: public, internal, confidential, and personal or legally protected.",
    "item": 1
   },
   {
    "say": "Then decide which level may go into which kind of tool. Public data can go into general-purpose tools, but confidential and personal data belong only in approved business plans or self-hosted setups.",
    "item": 2
   },
   {
    "say": "And always put in only the minimum data needed to finish the task.",
    "item": 3
   }
  ],
  "EP11 | Three Security Layers for Using AI at Work#6": [
   {
    "say": "Layer two is masking, also called de-identification.",
    "item": 0
   },
   {
    "say": "Replace names, phone numbers, emails, addresses, tax and account numbers, exact dollar amounts, and customer names with placeholders.",
    "item": 1
   },
   {
    "say": "Keep a lookup table inside your company, and don't store it in the same place as your AI conversations.",
    "item": 2
   },
   {
    "say": "When the AI gives you a result, swap the placeholders back for the real details in your internal environment.",
    "item": 3
   }
  ],
  "EP11 | Three Security Layers for Using AI at Work#7": [
   {
    "say": "Layer three is accounts, permissions, policy, and audits.",
    "item": 0
   },
   {
    "say": "Use company-managed accounts or business plans, and never personal accounts for company data.",
    "item": 1
   },
   {
    "say": "Set permissions by role, and disable accounts when people leave.",
    "item": 2
   },
   {
    "say": "Review each service's data-control options, like whether your data may be used to improve the service. Feature names and availability vary by version and region, so check the official documentation.",
    "item": 3
   },
   {
    "say": "Finally, write an AI use policy covering approved tools, data that must never be entered, when human review is required, and how to report an incident.",
    "item": 3
   }
  ],
  "EP12 | From Resistance to Company-Wide Use: A Three-Step Rollout#1": [
   {
    "say": "Now episode twelve, a three-step rollout. Here's how to shoot the video.",
    "item": 0
   },
   {
    "say": "The hook asks: bought a pile of AI accounts and nobody uses them? The problem isn't the tools.",
    "item": 1
   },
   {
    "say": "The pain point is that the owner buys the tools and books a seminar, but employees say it's not useful, or it's too much trouble.",
    "item": 2
   },
   {
    "say": "The hands-on part shows three steps: pick a small, repetitive use case for a Quick Win, build shared resources, and deal with the people side.",
    "item": 3
   }
  ],
  "EP12 | From Resistance to Company-Wide Use: A Three-Step Rollout#2": [
   {
    "say": "To close the video, the message is to start with one small use case so people feel the benefit first.",
    "item": 0
   },
   {
    "say": "Then you invite viewers to comment to get a rollout checklist.",
    "item": 1
   }
  ],
  "EP12 | From Resistance to Company-Wide Use: A Three-Step Rollout#3": [
   {
    "say": "Again, the hook needs a small correction.",
    "item": 0
   },
   {
    "say": "The three blind spots in the video are a way to organize the episode, not a universal law. Rollouts fail for different reasons in different companies.",
    "item": 1
   },
   {
    "say": "Common causes include no clear use case, no time to practice, fear of being replaced, unclear security rules, and managers who don't use AI themselves.",
    "item": 1
   },
   {
    "say": "What they share is that buying a tool gets treated as adoption, when real adoption means changing how people work.",
    "item": 1
   }
  ],
  "EP12 | From Resistance to Company-Wide Use: A Three-Step Rollout#4": [
   {
    "say": "This one is for managers, HR, department heads, and anyone leading digital transformation.",
    "item": 0
   },
   {
    "say": "You don't need a big team. It works for teams of fewer than ten people too.",
    "item": 1
   }
  ],
  "EP12 | From Resistance to Company-Wide Use: A Three-Step Rollout#5": [
   {
    "say": "Step one is picking the right use case for a Quick Win.",
    "item": 0
   },
   {
    "say": "Look for work that's highly repetitive, has clear rules, produces easy-to-check results, and costs little if it goes wrong. Think meeting notes, support FAQ drafts, first drafts of social posts, or quote formatting.",
    "item": 1
   },
   {
    "say": "Avoid starting with a high-risk process, or one that spans many departments.",
    "item": 2
   },
   {
    "say": "And write down success criteria: how long it takes each week now, how much shorter you want it, and how you'll check quality.",
    "item": 3
   }
  ],
  "EP12 | From Resistance to Company-Wide Use: A Three-Step Rollout#6": [
   {
    "say": "Step two is building shared resources to lower the barrier to entry.",
    "item": 0
   },
   {
    "say": "Start an internal shared prompt library. Each entry should have its purpose, a usage example, cautions, who maintains it, and a version date.",
    "item": 1
   },
   {
    "say": "Share success stories from different departments, not just tool introductions.",
    "item": 2
   },
   {
    "say": "And name seed users in each department, people coworkers can ask when they get stuck.",
    "item": 3
   }
  ],
  "EP12 | From Resistance to Company-Wide Use: A Three-Step Rollout#7": [
   {
    "say": "Step three is dealing with the people side, not just the tool side.",
    "item": 0
   },
   {
    "say": "Communicate honestly. AI should take on repetitive work so people can focus on judgment and relationships. Don't promise that no one will be affected.",
    "item": 1
   },
   {
    "say": "Instead, explain your approach: how hours will be reallocated, what training is offered, and the principles for role changes.",
    "item": 1
   },
   {
    "say": "Give people practice time and a safe place to make mistakes.",
    "item": 2
   },
   {
    "say": "Put the time saved toward more valuable work. Otherwise, people feel they're just being asked to do more.",
    "item": 3
   },
   {
    "say": "And managers, lead by example. Use AI yourself.",
    "item": 4
   }
  ],
  "Course Recap and Next Steps#1": [
   {
    "say": "Let's recap all twelve episodes, starting with marketing.",
    "item": 0
   },
   {
    "say": "Episode one, social posts: build a brand voice library first, adapt one topic across platforms, and never skip the review.",
    "item": 1
   },
   {
    "say": "Episode two, product scene images: never alter the product itself, use the real product as the reference, and run A B tests.",
    "item": 2
   },
   {
    "say": "Episode three, multilingual avatars: get consent, label content as AI-generated, follow platform rules, and have a human proofread.",
    "item": 3
   }
  ],
  "Course Recap and Next Steps#2": [
   {
    "say": "Next, internal operations and admin automation.",
    "item": 0
   },
   {
    "say": "Episode four, meeting notes: require source text as evidence, and have a human confirm owners and deadlines.",
    "item": 1
   },
   {
    "say": "Episode five, contract review: AI assists but doesn't give legal advice, and major contracts still need a lawyer.",
    "item": 2
   },
   {
    "say": "Episode six, automated workflows: separate automatic steps from human-confirmed ones, protect your API keys, and design for errors.",
    "item": 3
   }
  ],
  "Course Recap and Next Steps#3": [
   {
    "say": "Then sales and customer service, and analysis.",
    "item": 0
   },
   {
    "say": "Episode seven, support chatbots: maintain the knowledge base, hand off to a human when it can't find an answer, and test before launch.",
    "item": 1
   },
   {
    "say": "Episode eight, cold outreach: use legitimate sources, include real recent news, offer an unsubscribe option, and never abuse it.",
    "item": 2
   },
   {
    "say": "Episode nine, data analysis: mask first, inspect next, analyze last, and always verify the math.",
    "item": 3
   }
  ],
  "Course Recap and Next Steps#4": [
   {
    "say": "And the final group of episodes.",
    "item": 0
   },
   {
    "say": "Episode ten, competitor research: every number needs a source and a date, avoid hallucinations, and use only public information.",
    "item": 1
   },
   {
    "say": "Episode eleven, security: classify your data, mask it, manage accounts and audits, and verify each service's current data-use terms.",
    "item": 2
   },
   {
    "say": "Episode twelve, adoption: pick a Quick Win, build a shared library, and deal with the people side.",
    "item": 3
   }
  ],
  "Course Recap and Next Steps#5": [
   {
    "say": "Across all twelve episodes, five principles keep coming back.",
    "item": 0
   },
   {
    "say": "One: humans make the judgment, AI writes the draft. AI output is a first draft, not the final answer.",
    "item": 1
   },
   {
    "say": "Two: give facts first, then ask for output. Limiting the data scope reduces fabrication.",
    "item": 2
   },
   {
    "say": "Three: sources make it credible, so use citations, dates, and spot-checked calculations.",
    "item": 3
   }
  ],
  "Course Recap and Next Steps#6": [
   {
    "say": "Here are the last two principles.",
    "item": 0
   },
   {
    "say": "Four: minimum data, only when necessary, with masking, permissions, and audits.",
    "item": 1
   },
   {
    "say": "Five: start small, then scale. Validate with a small use case, then expand.",
    "item": 2
   }
  ],
  "Course Recap and Next Steps#7": [
   {
    "say": "One last thing: how do you know your AI adoption is working? Use an estimate, not a feeling.",
    "item": 0
   },
   {
    "say": "Hours saved is the original hours per task, minus the hours after adoption including review, times the number of tasks per month.",
    "item": 1
   },
   {
    "say": "Cost saved is those hours times your hourly staff cost, which you estimate yourself, minus tool and maintenance costs.",
    "item": 2
   },
   {
    "say": "And track quality and risk at the same time, so you don't lose accuracy just to go faster.",
    "item": 3
   }
  ],
  "closing": [
   {
    "say": "That's the end of this chapter, and the end of the course. Please do the lab and the chapter quiz to check what you've learned.",
    "item": 1
   },
   {
    "say": "Then take the final review. Remember the five themes: human judgment, facts first, sources, minimum data, and starting small. Thanks for learning with me.",
    "item": 1,
    "pause": 1
   }
  ]
 }
}