🤖 AI Customer Support: Automate Repetitive Tickets Without Losing Customer Trust

Speaker A: Welcome to Marketing in the age of A.I. uh, the show for leaders who want real results, not more noise. Join. Emmanuel rose as he helps business builders turn AI Uh from a confusing add on into a practical advantage with clearer messaging, stronger trust, and smarter systems. If you felt the shift happening in real time, you're not imagining it in the agentic pivot. Emmanuel calls it the move from experimenting with AI to building with it. So it supports your team, your brand and your growth. Let's get into it.

Speaker B: Here's a number that stuck with me this week. One company put AI on its customer support and saved 1700 hours in the first month. That was just the first month. Another company replaced about 700 support people with a bot. A year later, the CEO said, I'm quoting him, we weren't. We went too far. Same technology, two completely different outcomes. The difference wasn't the AI, it was what each one did with the humans. Right? That's the whole show today. Human in the loop. Now, I've hit, uh, support bots like you would probably have, and they sometimes they are super helpful, sometimes they are moderately helpful, and sometimes they are not helpful at all. And it definitely impacts my, uh, for me, if I hit a bot that is not helpful, then it has a negative impact for how I, I view that brand. Be interested to hear your thoughts on that also. All right, well, welcome to Marketing in the Age of AI. One location here in eastern Oregon. And thanks for spending a little bit of time with me today. Today we're talking about AI and customer support. And, um, let me say what that actually means in plain words. It's software that reads a customer's question and answers it on its own. Think of it as a really good front desk person who handles all the easy stuff. So the manager only gets pulled in the hard calls, same questions. The AI just closes the tricky ones it hands to a human. And if you're in marketing, not support, stay with me. Support is where your brand actually gets tested. Every bot that frustrates a customer quietly undoes the trust. Your brand and your marketing paid good money to build. The lesson carries into everything else you sell. Here's where we're going. Four stops. First, the news, including the big platforms you already pay for quietly shipping. Then, my hot take on the companies racing to automate everything. Then the players of money in a real case study where AI actually worked. So let's get into it. Let's start with what moved this week. First, a company called Hiver launched a new platform on October 12th. It's built around what people are calling AI agents. Let me define that, because you'll hear it all day, all week, and for the future. An AI agent is software that can take a few steps on its own to finish a job, not just hand you one answer and stop. So instead of a bot that reads you a help article, the agent actually digs into your order, checks the system, and tries to fix the issue you're having. Hiver's pitch is that the agent does the digging and the human makes the call. That's the model to watch, not a robot replacing your team. A robot doing the grunt work so your team can think. Second, Zendesk. They're one of the big names your support team probably or could already use. They rolled out what they call specialized AI agents built for a specific industry, starting with online stores, returns, refunds. Where's my order? What's my tracking number? All of it. Here's the part that matters for you. You can now build one of these agents without writing any code. You can configure it to the way you'd like to train a new hire. They claim it can automate up to 80% of the workflow. Hold on to that number. Up to 80% is a ceiling, not a promise, and I'll come back to it. Third, and quickly, because it's plumbing, Cisco, the company behind a lot of your office phone and video gear, is adding AI agents right into WebEx. The plain takeaway is this. The tools you already pay for are adding AI support for free inside their next update. So before you buy anything new, check out what you've already got. And one last thing, because it's proof, not hype. An independent research firm, Nucleus, studied reels and desk AI customers. Not the vendor talking, the customers. They found teams hit real, measurable value in about 25 days. Admin saved about 5.5 hours a week. And customer satisfaction scores actually went up, not down. So here's what to do with all this. Before you shop, ask your current support software one question. What is a already? What? Uh, excuse me, what? AI is already included in my plan, and you may already be paying for it. Here's my hot take of the week. The goal is not to replace your support team. The goal is to delete the repetitive tickets so your humans can handle the ones that actually need a human. Remember that 80% number. Here's the honest version. Out in the real world, these tools close somewhere around half to two thirds of questions on their own. The rest need a person. When a vendor leads with up to 80%. Ask them for the average, not the ceiling. Because here's what happens when you believe the ceiling. You fire the humans, you point the bot at everything. And then the hard, emotional this is my money questions hit a wall. That's the Klarna story. They swapped out around 700 people for a bot. For a while, the numbers looked amazing. Then the quality cracked and the CEO admitted they went too far and they started rehiring. So measure two things, not one. Measure how many tickets the AI handles and measure whether your customers are happier or angrier. Most teams only watch the first number, celebrate the ticket drop and never notice. They just taught their best customers to hate calling in. Delete the tickets, keep the people. That seems to be the game right now. Key companies and tools. Let me give you the lay of the land, because the list is long and most of it sounds the same. Three buckets. First, the incumbents, the safe, sticky layer. Zendesk and Intercom's fin agent, uh, live right inside your help desk. And your help desk. That's just the inbox and ticket system for your support team, where they already work all day. These win because they already have your data and your team's habits. The AI is one more button, not a whole new tool to learn. One term here, because it's the only number that matters when you shop. Resolution rate. That's the share of questions the AI closes without a human ever stepping in. Ask every vendor for it. Real ones land somewhere around 55 to 80%. Second bucket, the Enterprise agents, Decagon and Sierra. These are powerful, deeply custom, and honestly priced for big companies. For instance, these run from roughly 95,000 a year to up to hundreds of thousands. That's not in your budget. That's fine. That's not who they're for. Third, and this is the one for team like teams like yours. The challengers to sit on top of what you already have. For instance, a tool called Easel AI plugs into your existing help desk without ripping anything out. It charges about $0.40 per ticket instead of the big yearly contract. So who's winning and who's exposed? The winners own your data and your workflow, stuff you don't want to rebuild. Exposed ones are, uh, the thin bots with no connection to your real systems. And any team measuring ticket deflection while ignoring whether the customers are happy. Pick the tool that fits your size. Then watch the resolution rate and the sentiment together. Follow the money. Let's follow the money, because that shows you where it gets real. Three moves and I'll Lead each one with you three moves and I'll lead each one with what the company actually does for a business like yours. 1. Decagon builds custom AI support agents for big companies. This week it was reported they're in early talks to be bought outright by the service giant ServiceNow. An acquisition that just means one company buying another whole company. They've raised about $481 million and they're valued around 4 1/2 B valuation just means what investors think the whole company is worth. The signal when a giant would rather buy than build. The category is now core, not a, uh, nice to have. 2. Parloa helps big companies run whole fleets of voice and chat agents working right alongside their human reps. Note that alongside the humans, not instead of them. They raised $350 million earlier this year, what's called a series D, which is just a later round of investor funding. The money is flowing to platforms that manage people and AI together. 3. A newer company called Wonderful Builds. Wonderful builds AI service agents and it hit a $2 billion valuation on $150 million round. The point isn't the big number. Point is how much capital believes this is a lasting market, not a passive feature. Here's the operator. Read of it all the smart money is not betting on firing your whole team. It's betting on tools that keep humans in the loop and take the repetitive, time consuming work right off their plates. That tells you where this is heading. I think. Now the case study, the part where I show you this actually works for a company like yours, Anthropic. They build AI themselves. So when they needed to fix their own customer support, you'd think they'd build their own. They didn't. They bought a tool, Intercom's fin agent, and pointed at their support inbox. Here's what happened. In the first month, that AI was involved in almost every customer conversation and it fully closed half of them on its own. That saved 1700 hours in one month. A year in, it's closing 58% of the conversations it touches. But here's the part that I care about. They didn't fire anyone. They took all those freed up hours and moved their people onto the hard stuff. The complex enterprise problems, the tricky compliance questions. The kind of conversation where a frustrated customer needs a real human with judgment. The repetitive manual tickets, the ones that nobody liked doing anyway. The AI took those away. Now look, that's a software company, but if you run a 40 person SAS, a consulting firm, a wellness brand with a supporting box, you've Got the same split after your tickets are the same five questions over and over. A result like that is within reach. A team your size. So here's my first offer. If you're staring at a support inbox wondering where I AI even fits in, reach out to me, find me on LinkedIn or email me manualstrategicmarketing.com no pitch, no pressure. Just bring me the situation, uh, and I'll point you at a first step. That's it. Practitioner playbook. Here's your workflow for the week and you need zero AI experience to start. The whole first step is just reading your own tickets or your own, uh, inbox and needs from your customers. Step one, export your last 90 days of messages from your clients or customers. Open ChatGPT. Hopefully not the free version, but that will work. Also paste them in and type this Group these into the top 10 question types and rank them by how often they come up. For most teams, the top three are 60 to 70% of everything. That's your target. Step two, sort those top types into three piles. Questions already answered in your help documents, questions that need a human expert, and question with no written answer anywhere. Only the first pile is ready for AI today. The third pile tells you what to go and document. Step three, turn on your AI in co pilot mode first. That means it drafts the answers a human hits send. You watch for a couple weeks before you let it reply on its own. Step four, track the resolution rate. A good target is 40% close on its own in 30 days, 60% at 90 days. And watch your satisfaction scores the whole time. If the number of tickets drops but customers get grumpier, you're going to need to slow down. That's it. Read your tickets. Start small. Keep a human in the loop. All right, well, let's bring it home. If you take one thing from today, take this. Don't automate your people out. Automate the boring tickets so your people can do work only a human can do. Delete the tickets. Keep the humans Quick one before you go today. Today you got today's show got you thinking about writing down how your team actually handles those repeat questions and that's step one. Before any AI can help, take a look at Strategic sop. It's a tool for documenting and systematizing your team's workflows so the knowledge lives somewhere besides people's heads. It's for operation leaders and growing teams getting the processes out of their heads and on the paper. Find it@strasticsop.com and this is where you come in. Send me your question on LinkedIn and might answer it on a future show. Thanks for spending some time with me today and I look forward to seeing you again real soon.

Speaker C: Thanks for listening to Marketing in the Age of AI. Be sure to subscribe wherever you heard this podcast so you never miss a future episode. Your copy of Authentic Marketing in the age of AI is available at Amazon.com or emmanuelrose.com you can also connect with Emanuel on LinkedIn. Thanks again for listening and we'll see you next time.