How B2B Brands Use AI for Real-Time Sales Coaching

Lucas: Luna, I want to talk about something that's quietly reshaping how B2B sales teams operate - real-time AI coaching during live calls. Luna: You mean like an AI agent that listens in and whispers suggestions to the rep while they're talking to a prospect? Lucas: Exactly. It's not just recording calls for later review anymore. The AI is there in the moment, analyzing sentiment, detecting objections, and surfacing talking points. And it's moving fast - about 40 percent of enterprise sales teams now use some form of AI coaching tool, up from 12 percent just two years ago. Luna: That's a huge jump. What's driving the adoption? Lucas: A few things. First, the cost of large language models has dropped significantly since 2024, so startups can build this stuff without a fortune. Second, the remote and hybrid work reality means managers aren't sitting next to reps anymore. They need a different way to coach. Luna: So it's partly a remote work solution. But does it actually work? I've heard reps complain that having an AI interrupt them mid-sentence is more distracting than helpful. Lucas: That's a real concern, and early versions definitely had that problem. But the better systems are designed to be non-intrusive. They don't interrupt. Instead, they show a subtle prompt on the rep's screen - like a key objection that just came up, or a reminder to ask about budget. The rep decides whether to use it. Luna: So it's more like a live cheat sheet than a coach barking orders. Lucas: Right. And the data is compelling. I looked at a case study from a mid-market SaaS company - about 200 employees, selling a cybersecurity product. They deployed a real-time coaching tool from a vendor called Second Nature, and within six months, their close rates were up 30 percent. Average deal size also increased by 18 percent. Luna: Thirty percent is not a small number. But was that purely the AI, or did they change other processes too? Lucas: They did have a parallel training initiative, but the company attributed most of the improvement to real-time prompts. Specifically, the AI flagged when a rep was about to go off script or when the prospect showed buying signals. The reps said it helped them stay focused and not miss opportunities. Luna: Who else is playing in this space besides Second Nature? I know Gong and Chorus have been adding features. Lucas: Gong launched a real-time coaching module in early 2025 called 'Gong Live.' It uses their massive dataset of recorded calls to predict what works. Chorus, now part of ZoomInfo, has a similar feature. And there are newer entrants like Wingman and Jiminny that were built for this from the ground up. Luna: What about smaller teams that can't afford those enterprise platforms? Any options? Lucas: Some of the newer tools are surprisingly affordable. Wingman, for example, starts at about $30 per user per month. And there's an open-source project called Otter Coach that lets you build a basic version using OpenAI's API. You don't need a huge budget to experiment. Luna: That's encouraging. Let's talk about the technology a bit. How does the AI know what to prompt? Is it just keyword detection? Lucas: It's more sophisticated now. They use a combination of speech to text, sentiment analysis, and large language models that understand context. For example, if the prospect says 'we're happy with our current vendor,' the AI recognizes that as a competitive objection and might suggest a comparison framework. It's not just keywords - it's understanding the flow of the conversation. Luna: But the model has to be trained on good sales conversations to begin with. Garbage in, garbage out, right? Lucas: Absolutely. That's why the early adopters tend to be companies with a large library of recorded calls. They can fine-tune the model on their own top performers. But even without that, the pre-trained models from companies like Gong are surprisingly effective because they've been trained on millions of calls across industries. Luna: What about compliance? If you're recording and analyzing calls in real time, you need consent in many jurisdictions. Lucas: That's a critical point. Most tools handle it by requiring that both parties are notified at the start of the call. The AI doesn't kick in until consent is confirmed. And the data is usually encrypted and stored in a way that's compliant with GDPR and CCPA. But it's something every buyer needs to vet carefully. Luna: Let me ask the elephant-in-the-room question: does this replace sales managers? Or does it change their role? Lucas: It definitely changes their role. The manager shifts from being a 'gotcha' reviewer - listening to recordings to find mistakes - to being a strategist. The AI handles the real-time tactical coaching, and the manager focuses on skills development, pipeline strategy, and career growth. Companies that have done this well report higher manager satisfaction too. Luna: So the AI is more of an enabler than a replacement. That makes sense. Lucas: Exactly. And the reps I've talked to actually prefer it. They get immediate feedback instead of waiting a week for a call review. It's like having a co-pilot who's seen every deal ever closed and can whisper the playbook at the right moment. Luna: What's the biggest risk? If a rep leans too heavily on the AI, do they lose their ability to think on their feet? Lucas: That's a valid concern. Some organizations have reported that junior reps become too reliant on the prompts and struggle when the system goes down. The best practice is to use the AI as a training wheel - gradually reduce the frequency of prompts as the rep gains experience. And always pair it with traditional role-play and feedback. Luna: So it's not a magic bullet. But when done right, it seems like a force multiplier. Lucas: Exactly. And I think we're still in the early innings. As voice models improve and latency drops, the AI will be able to analyze tone and pace, not just words. Imagine a tool that tells you 'the prospect's voice just became more hesitant - slow down and ask an open-ended question.' That's coming. Luna: That's both exciting and a little eerie. But I can see how it would help reps who struggle with reading the room, especially over video. Lucas: Before we wrap up, I want to say something quickly. If you're finding these conversations useful in your own work, consider supporting the show at buy me a coffee dot com slash fexingo. It's what keeps us ad-free and focused on topics like this. No pressure, just if it's added value for you. Luna: Yeah, it really does help us keep the lights on and continue digging into the stuff that matters for revenue teams. Lucas: So back to the future - I think the next frontier is integrating real-time coaching with CRM data. Imagine the AI not only suggesting what to say but also pulling up the prospect's history and predicting their next likely objection based on similar deals in the pipeline. Luna: That would be incredibly powerful. It turns the sales call into a data-driven conversation, not just a pitch. Lucas: Right. And that's the real promise of AI in sales - not replacing the human connection, but making it more informed and effective. Thanks for listening, and we'll see you next time on The Growth Operator.