Loading summary
Host
We have something exciting. Mira Moradi, who was the former CTO of OpenAI, started this new company, Thinking Machines.
Co-host
You ready?
Host
And they just had their big first announcement and it's all about voice. AI.
Tech Expert
Absolutely. I'm ready. What's up?
Host
Here's a live demo from Thinking Machines.
Co-host
Yeah, so we're giving an announcement today and I've got two of my friends coming to help. Every time one of them enters the frame, I need you to say friend.
Tech Expert
Got it. I'll say friend whenever one of them walks in.
Host
So this is having the AI use video and voice in real time to take some action. Yeah.
Co-host
You can stream input into it in real time and it can respond to you even while you're speaking to it simultaneously. How does that sound?
Tech Expert
Sounds like a solid setup. Full duplex with real time interaction is super useful. Friend.
Host
All right, so you see the model using the video, it knew that the guy walked in, follow the set of instructions, friends. So you can imagine doing the same thing with saying hey, when somebody calls with this question, respond in this way. You could see that voice can become a much more proactive and live integrated. And so the big trend and what Mira and Thinking Machines team have done is introduced a concept that real time voice is going to be one of the next big frontiers in AI. But to make it interesting, I'll do it in Hindi. Can you translate to in English in
Co-host
real time for my friend and for audience?
Tech Expert
Absolutely. I'll translate as you go. Today we're taking a look at our preview model to release it.
Host
This is going to be really transformative for how people work. The multilingual side of this is super fascinating to get Hindi to English in real time from a translation perspective. Fascinating. And then again, it's going to say friend like it just did because you had a second person walk in. And so it's showing you the power of the model to be able to interact in real time with this context from the video. This is going to be a next huge shift in AI. We're going from this chat bot to real time audio that can actually interject in real time, translate in real time, assess what's happening in video real time. It's going to unlock the ability to have brand new experiences and to solve problems in new and different ways. The hottest trend in AI right now isn't video, it isn't agents, it's voice. And that's because there have been two huge foundational technology breakthroughs that are going to shape the way you, your business, your brand show up through AI through voice. And your brand is about to get a literal voice online. And that is going to be really powerful. The first big thing that's happened is that OpenAI just a few days ago released a new Update. It's called GPT Real Time 2 Voice. And what's going to be much better than me explaining this is to show you a quick video of how this works.
OpenAI Representative
OpenAI just pushed Voice AI into a completely different level. And the biggest change is not only that the AI sounds better, but the real change is how naturally it can now hold conversations real big time.
Tech Expert
Let me pull the latest context and update your CRM. Sablecrest launched warehouse automation this morning.
OpenAI Representative
Earlier, voice assistants often felt slow and awkward.
Host
So the big change here, the big gap that's been in place with AI voice, it's been that there's been a lot of latency in terms of going back and forth and doing voice in real time. We. Which is the big unlock, right? Think of all of the use cases, call centers, how you interact with technology, websites, et cetera. Look, this is one of those topics where the more you dig in, there's more to work with. So we put together a free resource that gets you all the specifics and covers a lot of what we didn't get into or have time to cover. Here you can click the link in the description or scan the QR code to get the full resources of how to adopt Voice AI for your company. OpenAI has made really big strides here. And the interesting thing about this, we had a big announcement from a company called Thinking Machines. Thinking Machines was founded by the former chief technology officer of OpenAI. Thinking machines have gone after and solved that big problem. I was just talking about the problem of latency. Yes, OpenAI has solved a lot of it, but Thinking Machines seems to be out of the gate even better at that latency problem. Let's take a quick look.
Demo Participant
I heard mountain biking is a lot of fun, so I wanted to go mountain bike and I was thinking to actually take my parents with me. They're pretty old. They're 70, almost 80. But I thought that would be a fun trip together.
Tech Expert
Wait, no, don't take them mountain biking at 80, that's incredibly risky. They could easily get seriously hurt. Pick something safer.
Host
I love that he's trying to take
Demo Participant
his parents mountain biking on a normal mountain bike. We can do like tandem bikes. Maybe they're more sturdy and we can find like, you know, some volcano or better like an active volcano.
Tech Expert
No, stop. An active volcano is deadly.
Host
Okay. He's trying to be hyperbolic here, taking his 70, 80 year old parents mountain biking or taking them to an active volcano. That's not the point. What the point is is that the latency, the speed of these models to react in real time has gotten so low that the model could interrupt him and in the flow of normal speech, as a human would go in, interject and say, hey, that's a really bad idea. What they're actually trying to demonstrate is that Voice AI is crossing a chasm from this like very reactive dictation mode and type of technology to very proactive technology that can be your partner, can interject, can make you better, just as if you were like having a chat based conversation. And this is especially valuable when maybe somebody's interacting with you on your website. From a marketing perspective, somebody needs a customer support ticket. Those types of things are going to be really transformed by this type of technology. And so thinking machines starting in Voice AI, I think, and this being their first big announcement is pretty wild. It's a real signal that there's going to be more and more competition and more competition is going to lead to more innovation for these fundamental technologies around Voice. You're like, cool. These two big startups have released this real time AI technology. What does this actually mean for me? What does it actually mean for my business? And there's a couple of things here, right? You saw these big announcements, which is great, but how it handles the pause, how it starts becoming part of the kind of normal way of working shows that it can actually become part of society. And anything that starts getting integrated into just how we work becomes a marketing opportunity, becomes something that marketers have to think about and have to decide, hey, am I going to participate? Am I going to be a part of this? And I'm bringing this up at a time when AI feels completely overwhelming, right? There's new AI stuff all the time and I'm not here to tell you that, oh, real time Voice, like that's so cool. And this is the new thing. What I'm saying is that this trend is just getting started. And what I'm doing is predicting that over the next couple of years you're going to see a mass increase in AI voice across all the ways you and your customers interact. And those become marketing opportunities. So if you're thinking out there, you're a founder, you're a marketer and you're like, how do I actually need to think about this trend? And capitalizing on this trend? There's a few things I'd recommend you do and not do. The first is you now have an actual brand voice. Like your actual brand has a tone. And we've talked about on past shows that brands are starting to create custom avatars, AI actors for their videos. You're going to need a custom voice that represents your brand and there's great tools to do that. You can do that in 11 labs and several other models. Maybe in a future show you can leave a comment. If you want us to show you how to create a custom brand voice, we can do that. But this is one thing that you're going to have to consider, and this is a really good example of the trap that comes up here. Technology like this, which is like voice APIs in real time, seems dorky. It seems like something you would have your IT person go and deal with. If you have your IT person, whether you're a CIO at a big company or you just have like an outsourced IT person at a smaller company, you're going to miss a lot of the marketing and growth benefits of this type of technology. The first thing, like I said, is you're not going to actually be able to communicate what your brand is through this voice. And there's going to be a lot of aspects. Is your brand warm? Is it friendly, what accent does it have? All of these things that within a few words you can communicate to somebody just through what that voice is in a way that like sometimes you only get a few words. Nowadays, attention is so scarce, so you do not want to miss the opportunity. And so the number one thing here is you have to start thinking of voice as a channel and a modality. And I think that's very different. I think when it comes to AI, most people I talk to are AI is text and AI is chatting in a tech spot or if you're a little advanced, building an agent. When a buyer asks AI for a solution like yours, does your business come up? Most companies have no idea. And by the time they find out, they've already lost the deal to someone who did. HubSpot AEO helps you show up in those moments with the right answers buyers are looking for before the first click, before the first form fill. That's the moment HubSpot AO is built for. Head to HubSpot.com aeo to try that tool for free today. What we're saying now is that chat is moving to voice, and voice is going to become a channel that customers expect to interact with me and prospective customers expect to interact with me. 1. The second is because it's a channel. I have to care about how my brand shows up in that channel and I have to know what my tone is, what my actual custom voice is, and show up in the way that I'm looking to make happen. Then there's also just basic instructions that you're going to need to give these voice bots, just like you might do in Claude or ChatGPT, where you're writing system prompts or custom sets of instructions. We've done experiments at HubSpot. People will talk to these voice agents for like 5, 10, 15 minutes. And in those times, you don't want that discussion to go widely off the rails. And so you need to keep that voice agent trained and focused on the task at hand. So custom instructions are going to be a big part of. As you start to test and roll out technology like this, the biggest thing here is you want people to understand that this is an AI interaction, but you want it to sound natural and human. You know, the technology is still developing. This is why this is not like you have to go do this all what I'm telling you immediately it says, wow, this is a big step forward in the technology and I think it's going to keep improving at a very exponential rate over the course of the next year. And, and what I'm trying to help you with, it's not wake up and say, oh gosh, I'm way behind and now I have to do all this voice stuff. It is stay on the trend and start building the foundation. Right? The biggest thing here as you're building this isn't that you have to have like a human sounding voice. The goal is to be about brand sounding, like we were talking about earlier. It has to represent your company, your brand, to just be a random voice agent or some random voice that's going to be okay, but that's not going to be what the best companies do. The best companies are going to make sure that their brand and who they are, their values come through in these new voice experiences. So I want to show you a third video in the show today. And this is about an example from a company, Sesame AI, which kind of hopefully brings this third point I'm making to life, which is it's about how do you make this voice engaging, representative of your brand, not just human sounding, but emotive and connected to the company you're building and the brand you're building. So we're going to get a quick demo here. Let's open it up.
Guest
Let me see if this will work.
Host
This week's AI for humans. We did a big chunk on it. And one thing I'll say before Alex is pulling this up is I. Oh, yeah.
Tech Expert
Long day for you or just getting started?
Host
I'm Maya.
Guest
Can you give us a second? I'm with Gavin Purcell. We're doing. Recording a podcast. He was just making a point and then we're going to. We'd like to chat with you, but
Host
Gavin, why don't you finish what you
Guest
were saying and then we'll pull Maya back in.
Host
Sure. So podcast, that's. So if the goal here is to have a voice experience that represents who your company is and what your brand is, and for those first seconds or minutes that somebody might interact with you in voice to be representative of who you really are in the company you're trying to be, well, then what do you actually do? Call your company. That's right. Does your company have a phone number? First of all, if it doesn't, that's problem one. And forget this show and go get yourself a phone number and a way to answer the phone. But call your company and does it have this old school robotic answering? Does it sound like your company at all? Is it a pleasant experience at all? If it's not, this is job one, fixing that basic interaction. One of the things that I've learned over time is because phone systems data are not like, as digital and as easily accessible in many cases as your website data and other sources of marketing, it often gets ignored and buried. So I would actually just go and just understand where the baseline is, understand how many people call that number, how do they get routed after they call that number? And then can I go call that number myself and get an actual understanding of what that experience is? For most companies, all likely it's going to be really bad. I did this a few years ago and it was literal gold. We had way more conversations coming in than we thought and we were not handling them in the best way possible. And so to continue to audit yourself and just know that there are new technology that could make these experiences way better. But before you even need to worry about the new technology, how are you doing? Like, what is actually happening? All right, so you're gonna get a baseline. Then, like, you have to start caring about it, you know, you have to then say, great, well, what should the experience be? What should this thing sound like? How should that routing actually work? If I was going to use one of these technologies, what is the strategy? What are the rules to make it work and make it work for me? And if you Believe this is a trend and if you're watching this far in the video, you probably do. You need some accountability. You need to make some progress. So today I want to close out with a couple of, I think fun and interesting predictions to close out the show. My first is within 12 months, an unknown brand is going to go viral for having a really awesome voice agent experience that's going to have fun personality, answer questions in a remarkable way and they're going to get millions, if not tens of millions of dollars of media exposure for doing this in a cool and interesting way. Could be you. The next is is there's going to be a category of voice first marketing and marketers that happen. And these early pioneers in voice are going to actually have a different surface area, different way of working than a lot of the tech space marketers. And there's going to be big opportunity. And if you're a marketer and you're looking to grow your career, voice could be a potential way to focus and differentiate and accelerate your career and your progress much faster. There's also there's always good, there's always bad. My third prediction is that there's probably going to be some wild lawsuit or scandal about a voice agent making a really bad mistake because these things keep happening and hallucinations and lack of clear instructions are still an issue for these models and what's happening in the world. And so this is what's going to also have a high risk of happening. If voice is lift solely to the IT team or technical team, you want it to be a partnership between the technical team and the marketer to make sure that these experiences are both accurate and and on brand. So here's my takeaway. It's not Voice AI is here. Everybody knows that Voice AI has been a bubbling trend for a while now. The takeaway is your brand is about to have an actual voice, a literal one. Customers are going to hear it more than they read your copy long term, I think. And if marketing isn't running, somebody else in your company is, you're not going to like how it sounds. And if it's going to be a big surface area for your customers, what are you doing as a marketer not being a part of that? So go make the call to your own company. Get that audit experience. I'd love to hear if you have any wild stories, post them in the comments below. Would love to hear what you learn in going through this experience, but I think it's the first step to actually learn what you should do with this new and rising category of Voice AI. Drop a comment with any future questions on Voice, hit like hit subscribe and we'll see you real soon on Market against the Grain I want to tell you about a podcast I love. It's called Nudge. It's hosted by Phil Agnew, it's brought to you by the HubSpot Podcast Network, the audio destination for business professionals and it's the UK's fastest growing business podcast. What I love about it is that the Nudge listeners love no fluff, no bs, evidence based marketing tactics they get in each episode. You're going to want to listen because this is like an MBA's worth of insight in every single podcast. And entrepreneurs. You're going to love the show because it's filled with repeatable, proven studies. Not hearsay, not one off success stories Marketers. You're going to love it because it discusses the psychology behind great marketing and what marketers are getting wrong. Listen to the Nudge wherever you get your podcasts.
Guest
This data is wrong every freaking time.
Tech Expert
Have you heard of HubSpot? HubSpot is a CF platform where everything is fully integrated.
Guest
Whoa. I can see the client's whole history. Calls, support, tickets, emails, and here's a task from 3 days ago I totally
Tech Expert
missed HubSpot grow better.
Marketing Against The Grain – "This Is the End of Chatbots"
Episode Date: May 14, 2026
Hosts: Kipp Bodnar (HubSpot CMO) & Kieran Flanagan (SVP Marketing)
In this lively and forward-looking episode, Kipp and Kieran dive deep into recent breakthroughs in AI-driven voice technology, exploring why marketers should care about the rise of real-time, proactive Voice AI—and how the evolution might signal the end for traditional chatbots. The episode centers on the disruptive announcements from Thinking Machines (a new voice AI company from OpenAI’s former CTO, Mira Moradi) and OpenAI’s own advances in natural, low-latency voice interaction. The hosts offer predictions on the future of voice in marketing, guidance for brands looking to ride this trend, and point out key pitfalls for companies who ignore this emerging channel.
“This is going to be a next huge shift in AI. We’re going from this chat bot to real time audio that can actually interject in real time, translate in real time, assess what’s happening in video real time.” — Host (01:34)
“The hottest trend in AI right now isn’t video, it isn’t agents, it’s voice.” — Host (01:52)
“The latency, the speed of these models to react in real time has gotten so low that the model could interrupt him and in the flow of normal speech, as a human would … What they’re actually trying to demonstrate is that Voice AI is crossing a chasm from this like very reactive dictation mode … to very proactive technology that can be your partner.” — Host (05:12)
Call Centers, Customer Support, & Marketing Applications (05:12 – 07:27)
Actionable Recommendations for Marketers (07:27 – 10:45)
“If you have your IT person, whether you’re a CIO at a big company or you just have like an outsourced IT person at a smaller company, you’re going to miss a lot of the marketing and growth benefits of this type of technology.” — Host (08:43)
“Call your company and does it have this old school robotic answering? Does it sound like your company at all? Is it a pleasant experience at all? If it’s not, this is job one … before you even worry about the new technology, how are you doing?” — Host (13:10)
“If voice is left solely to the IT team or technical team, you want it to be a partnership between the technical team and the marketer to make sure that these experiences are both accurate and on brand.” — Host (15:38)
“The takeaway is your brand is about to have an actual voice, a literal one. Customers are going to hear it more than they read your copy long term, I think. And if marketing isn’t running, somebody else in your company is, you’re not going to like how it sounds.” — Host (16:33)
Real-Time, Multilingual Voice Demo (01:21)
Voice is not just the new chat (09:10)
Future of Brand Engagement (16:33)
On Risk and Partnership (15:38)
Kipp and Kieran argue persuasively that the age of the text chatbot is winding down, being eclipsed by a wave of Voice AI breakthroughs that offer real-time, proactive, and context-aware interaction—ushering in an era where a "brand voice" is as literal as it is metaphorical. Their practical advice: marketers need to audit their current experiences, get ahead on creating custom branded voices, participate in shaping this new channel, and not leave the critical point of customer connection to tech teams alone.
If you’re a marketer, founder, or brand leader, now is the time to start thinking about—and investing in—your literal brand voice.