Latest / Intentional AI Daily / AI Vs. Humans In Customer Support
Transcript
- 0:00SPEAKER_00: What's happening, everyone?
- 0:00SPEAKER_00: Jason Wright here.
- 0:01SPEAKER_00: I've got Sarah here.
- 0:03SPEAKER_01: Hey everyone.
- 0:03SPEAKER_00: So I had a moment last week that I think every business owner needs to hear.
- 0:08SPEAKER_01: This sounds dramatic.
- 0:10SPEAKER_00: A client emailed me, frustrated, because an AI chat bot on a different company's site had given them completely wrong information about a refund policy.
- 0:19SPEAKER_01: Oh no.
- 0:20SPEAKER_00: They were ready to leave that company over it.
- 0:22SPEAKER_00: Years of being a customer gone because of one bad bot conversation.
- 0:26SPEAKER_01: That's rough.
- 0:27SPEAKER_01: So what's the takeaway?
- 0:28SPEAKER_01: Don't use AI in customer service.
- 0:30SPEAKER_00: No, the takeaway is knowing when AI should stay out of it.
- 0:34SPEAKER_00: There are moments where AI helps and moments where it actively damages the relationship.
- 0:38SPEAKER_01: Give me an example of each.
- 0:40SPEAKER_00: AI is great for the front door.
- 0:42SPEAKER_00: Answering common questions, scheduling and qualifying leads.
- 0:46SPEAKER_00: The kind of stuff that has clean answers.
- 0:48SPEAKER_01: Where does it fall apart?
- 0:49SPEAKER_00: Anywhere there's emotion, refunds, complaints, billing disputes, anything where the customer is already frustrated.
- 0:56SPEAKER_00: That's a moment where they need to feel heard, and a bot saying I understand your concern makes them more angry, not less.
- 1:02SPEAKER_01: I felt that.
- 1:03SPEAKER_01: Talking to a bot when you're already mad is the worst.
- 1:06SPEAKER_00: Exactly.
- 1:07SPEAKER_00: The bot is not actually understanding anything.
- 1:10SPEAKER_00: The customer knows that.
- 1:12SPEAKER_00: So instead of solving the problem, you've just told them you don't care enough to put a human on it.
- 1:17SPEAKER_01: So what's the rule?
- 1:18SPEAKER_00: Easy stuff, AI.
- 1:20SPEAKER_00: Emotional stuff, human.
- 1:22SPEAKER_00: Build the system so the handoff is fast and seamless.
- 1:25SPEAKER_00: The worst version is when you have to fight the bot for 10 minutes to even get to a person.
- 1:30SPEAKER_01: How does someone decide where their line is?
- 1:33SPEAKER_00: Look at what makes customers leave.
- 1:34SPEAKER_00: If your churn is happening at billing or support, that's where AI needs to step back.
- 1:39SPEAKER_00: If it's happening at first contact, AI can probably help.
- 1:44SPEAKER_01: Is there a middle ground like AI assists, but a human is still in the loop?
- 1:48SPEAKER_00: Yes, and that's actually where most of my client builds land.
- 1:51SPEAKER_00: The AI drafts the response, summarizes the issue, and pulls the customer history.
- 1:56SPEAKER_00: Then a human reviews it and sends it.
- 1:58SPEAKER_00: The customer gets a fast, accurate, human-signed reply.
- 2:01SPEAKER_00: Best of both worlds.
- 2:03SPEAKER_00: It's also way easier on the team.
- 2:05SPEAKER_00: They're not starting from a blank screen on every ticket.
- 2:08SPEAKER_00: They're editing, not writing.
- 2:10SPEAKER_01: What about businesses that don't have the volume to justify any of this?
- 2:14SPEAKER_00: Then you don't need it yet.
- 2:16SPEAKER_00: Don't bolt AI onto a problem you don't have.
- 2:18SPEAKER_00: If you're answering 10 emails a day yourself and it's working, leave it alone.
- 2:22SPEAKER_01: What's the warning sign that you've gone too far with AI?
- 2:25SPEAKER_00: Customer complaints about the bot itself.
- 2:28SPEAKER_00: The moment people are mentioning the chat bot in their feedback, you've crossed the line.
- 2:33SPEAKER_00: The tool should be invisible.
- 2:35SPEAKER_00: If they're naming it, it's getting in the way.
- 2:37SPEAKER_01: That's a good signal.
- 2:38SPEAKER_00: Another one is response quality dropping over time, AI tools update, prompts drift, and what was working six months ago might be giving weird answers now.
- 2:47SPEAKER_00: If you're not sampling outputs regularly, you don't know what your customers are actually seeing.
- 2:53SPEAKER_01: So you're saying audit your own bot.
- 2:55SPEAKER_00: Talk to it like a customer would.
- 2:57SPEAKER_00: Ask it the hard questions, see what it says.
- 3:00SPEAKER_00: If you'd be embarrassed by the answer, your customers are too.
- 3:05SPEAKER_01: Have you ever had to pull AI out of a workflow after the fact?
- 3:08SPEAKER_00: Once.
- 3:09SPEAKER_00: A client had AI handling appointment changes, and people kept getting double booked because the bot couldn't read context across messages.
- 3:17SPEAKER_00: We pulled it back to humans within a week.
- 3:19SPEAKER_00: Sometimes the right move is admitting it didn't work and reverting.
- 3:23SPEAKER_01: That's a real one.
- 3:24SPEAKER_00: Are you an entrepreneur or small business owner who has critical assets online like I do?
- 3:29SPEAKER_00: Your website, your email, your customer data, it's all sitting out there.
- 3:34SPEAKER_00: Here's the thing most people don't realize.
- 3:37SPEAKER_00: Your email and password are probably already floating around on a criminal marketplace, and you'd never know it.
- 3:43SPEAKER_00: Head over to digitalmafioso.ai and run the free scan.
- 3:47SPEAKER_00: Takes a couple of minutes, you'll see exactly what's exposed and what's vulnerable, plus you'll get clear options for fixing it.
- 3:54SPEAKER_00: Thanks for listening.
- 3:55SPEAKER_01: Catch you later.
- 3:56SPEAKER_00: See you in the next episode.
- 3:59SPEAKER_01: Thanks for tuning in.