HubSpot Breeze - Everything B2B firms need to know

Read the full transcript here:

Out of all the companies that you could have referenced earlier, why did Siemens spring to mind first? No comment. I think that says a lot about you. The world of B2B marketing, sales and service tech is moving fast. This podcast will come through the height of the hidden agendas and bullshit advice in search of the truth to help you make more informed decisions. Today we're going to talk about HubSpot Breeze. We want to take a very sort of honest...

or provide a very honest appraisal of Breeze. practical appraisal. Absolutely. There's a plethora of other reviews of Inbound that have been shared, blogs, podcasts, videos from other partners, and generally have played out the same storyline that HubSpot did at Inbound. A regurgitation of what? Yeah. I think we want to try and provide a bit of context as to why we think Breeze has come about.

Breeze as a set of features and functionalities that we've now had time to really play with. We've took some time to take our clients through some of the features and functionality, get their opinion, play with some of the workflows, some of the various different capabilities around AI as well. So think we're now in a more informed position, not just to regurgitate the messaging from HubSpot. We obviously got sight of Breeze coming in advance of inbound.

but as a partner under embargo until the conference itself. I think if we start by just giving people a quick flavor of what Breeze is, then we can talk about Breeze Intelligence particularly, because I think that's probably got the most relevance right now for B2B. We'll delve into why we think HubSpot has developed this particular part of its platform and what problems it addresses. And then we'll go into some of the key areas.

Do you want to kick us off by just talking about Breeze generally, just so people got a good understanding of what it's all about? Yeah, so Breeze is the name, the overarching name for the suite of AI-powered tools that HubSpot announced inbound a couple of weeks back now. It encompasses things that we'll go into in more detail later in the podcast, but agents, example. HubSpot have a new content agent.

(02:26.054)
a new customer agent, a new prospecting agent, and the fourth one.

Social. okay. And they are AI-based tools which guide you through a process of providing insight and then help you produce content within one of those four different areas in an AI kind of facilitated way, with the exception of the customer agent, which is more focused on AI chatbot capability. We'll talk a little bit more about that later. It also covers Breeze Intelligence, which we'll come on to, which is

really designed to help beef up the first party data that you have as an organization. And it's all powered by HubSpot's acquisition of Clearbit at the latter end of last year. memory serves me right? November, yeah. Which allows you to draw on a source of third party data to kind of backfill your own CRM and improve the quality of the data in your CRM.

And that also covers functionality such as form shortening and intent data, which we'll come onto in more detail. Okay. So it encompasses a lot of quite significant announcements from HubSpot a couple of weeks back, many of which are AI driven, some of which are very much focused on data. So I think our job today, staying true to the whole principle purpose of this podcast, Truth, Lies and B2B Tech, got it right this time. Lies, Truth and Tech.

lies and B2B. That's it. Spot on. Truth lies and B2B tech. Really, the whole purpose of this is to make sure that we're providing a nice take. We're a HubSpot partner. I've just got to say that. We're not going to blatantly dis HubSpot. And to be honest, I don't think it really warrants that in this particular instance. There's some great things that have come out of inbound. But HubSpot's fantastic at marketing itself. It was its conference.

(04:29.134)
face value you could get really really excited about some of the announcements and I think now we've had a little bit of time to really get under the skin of what's come out it's a it's important to say right maybe some things are further ahead yep and maybe other things there's kind of in that embryonic incubator stage then they'll definitely improve and they will do what it says on the tin at some point but maybe right now that this

Maybe a little bit of a shortcoming in certain areas. So I think if we start with breeze intelligence, yeah, because that for me is probably the most relevant for B2B right now. Other things I say may come later still need some more development and require an awful lot, particularly from an AI perspective, quality of data for AI to really do its stuff. The breeze intelligence for me, if we rewind to talk about

The context and maybe why it's all come about, let's explore the problem that we know a lot of organizations have in B2B. They have a whole mismatch of different data sources, all feeding its CRM, running through its CRM. So you've got first party data, second party data, third party data. Let's just explain what those different forms of data are. So first party data, that's the data that you collect.

yourself as a company through your website, through forms, through downloads and so on. Second party data is essentially data that you may have acquired through a partner or you may have acquired through a joint collaboration with a publishing house that's running a webinar series and you've got opt-in data that way. And then you have third party data, which is data you buy in from the likes of Zoom Info, from Cognizant or from Upleat and so on. And obviously from Clearbit.

part of HubSpot. obviously those different types of data, there's a spectrum of quality, of accuracy and of scale that goes across all three. So first party data, obviously you're going to get a higher quality, you're going to get higher accuracy because it's user submitted most of the time and it's collected directly from you. Collected at source, Second party data, you're going to again some quality, you're going to get

(06:49.742)
More accuracy because again, it's it might not be submitted directly to you But it's provided to you and but you're not going to get the same scale Third-party data is almost the reverse of that isn't it? You know, ultimately we've got very high volumes of data, but lacking in accuracy often lacking in Quality overall we normally say it's about what 60 to 70 cent if you're lucky in terms of any purchase data from a third-party source because he said

extensive process for them to repeatedly go around the same companies and validate data, people leaving all the time, changing roles, et cetera. it's a challenge for the third party data providers to keep on top of it. But we all know the quality of data that comes back from those various different sources. And therefore, I think there's other things that are happening as well in the world of data that do mean that arguably going forward, as it always has been first party data,

is the Nirvana. the most difficult to gather. But the most difficult to gather at scale. But there's other things at play now that would suggest that now to really function, to really be effective as a marketing operation as we move into 2025 and beyond, you need really clean data. Particularly if you're to start leveraging AI. Particularly you're going to start using AI to scale personalization. With poor data, you're going to make a real

mess of things. And Breeze Intelligence addresses that gap. Yeah. So let's talk about Breeze and what it, intelligence and what it does. So in simple terms, Breeze Intelligence helps keep your CRM database up to date. At a technical level, the way that it works is that within HubSpot Breeze Intelligence data enrichment tool, you can essentially ask the system to, and you can configure this in a way that works for you, but you can ask the system to continually enrich the data that you have in your CRM.

So for example, you have 10 records, so you're going to have many more. And on six of those, you have job title and on four of them, you don't. And Breeze Intelligence learns the job title of someone in your system, then that data can be automatically brought into your system so that it's as rich as possible. The data in your system is rich as possible, which as you've talked about is so important for AI because

(09:13.198)
particularly, and we'll come on to this conversation later, but particularly when our AI is embedded within your tools as Breeze is, and it's functioning based upon its understanding of your data. And of course, the greater richness, the greater quality of output you're going to get. So that's the first thing. It also helps maintain data that you already have. So if you have a job title in against a contact record and that changes or its view of the...

the data changes, then it can update that record for you. Or you can enrich on a company or contact basis in a manual fashion. And the use case for that is really sales. if, say for example, you have a very light record, it's not rich at all, you've got an email address and a first name and last name and maybe a job title. You can enrich that record manually if you were then to say, go and prospect that record.

You you wanted to find out more information about a company like revenue, for example, and you didn't have that. Breeze Intelligence allows you to manually enrich individual records in that way. can do it at a specific company and contact level. You can do it at bulk level as well. You can do it at an automatic bulk level so that your CRM database, your smart CRM is essentially being continually enriched, which is so important, obviously, with AI. And you pay for each time you enrich your data, don't you? Yes.

So it's a credit space system. And I think where some of the analysis still needs to happen, and of course it's dependent on your configuration, is what level of, basically what level of tokens, what level of credits you need in order to facilitate the different configurations that you've got. Of course you're going to need a much higher level of tokens if you want to continually enrich. But if you want to do, say for example, use form shortening, which I'll talk about in a second.

or you want to do enrich on a company by company basis, then you can certainly get away with less. So it really depends on your configuration, that one. But I've just touched on it there and form shortening is another part of Breeze Intelligence, which I think could be a potential game changer, particularly when it comes to conversion. Essentially, the concept of form shortening is that, say you had a form on your website and it had first name, last name, email address, city and job title.

(11:36.926)
and Breeze Intelligence knew four of those fields for that record and they weren't in your CRM, it can automatically shorten the form. So it doesn't show the ones that Breeze Intelligence already knows about. And so the contact, the prospect might only have to fill one field, but you still get all the data importantly. So it's helped with that ever present challenge of, okay, did we go for a long form and get more data? Do we go for a short form and get...

less data but more conversions. Essentially, you get the best of both worlds in that. So you're improving conversions and you're obviously improving the volume of first party contact data that you're... With no sacrifice on either side in an ideal world. So form shortening as a thing sounds a little bit, know, nodding, a bit basic, but actually can be quite impactful, particularly for companies that really want to improve conversion on their website and who doesn't want to do that? It sounds like a very tactical

part of the tool, but if you look at the wider impact that it could have, particularly these days where there's a lot of reluctance from people to share their data, even more so than ever before, conversions on forms, decreasing across the board. I there's a general skepticism about data privacy and data protection. so reducing the feeling that the

Prospect to the customer or whoever you're trying to target is having to hand over all of their personal information is going to increase conversion at a time when conversion rates falling as well. Yeah. Yeah. Well, we all know how off putting a long form can be, but ultimately a completely different topic. Not for today around gated non-gated content, but we can address that another time. so the third element of Breeze Intelligence, buyer intent, which really got me excited when I was

watching inbound and probably as excited as some of the speakers at the conference. I think I'd say a bit over excited. Maybe. Yeah. You saw you bore the brunt of that with the constant feed through slack. I think why I was excited because and I think this is really, really important that we explain this in the, in, you know, in the right terms and compare it to what other companies and what other people might regard buyer intent.

(14:01.976)
to be. our first exposure to buyer intent in any real meaningful way was with Zoom Info a few years ago. So let's just explain how buyer intent works with Zoom Info. So Zoom Info has a method of collecting data from thousands and thousands of different websites, most of the high authority sites. It will look at how people are interacting with certain topics, what

content they're downloading, et cetera. And then it will provide signals based on surges and spikes of interest in a particular topic and relate that back to the company where that surge has come from. Really, really powerful stuff. In theory, we've used it mixed bag of results in truth, but so that's one form of buyer intent. Now Sixth Sense also have a buyer intent capability.

And it's probably somewhere in between what HubSpot do and what ZoomInfo do. So as I understand it, they collect data, again, from across the internet. They amalgamate that, aggregate that with your own data that's collected from your website and maybe your CRM. And it will look to then go further than ZoomInfo and you'll start to pinpoint potential buyers in a certain process and what stage they're at, et cetera. And also map on ideal prospects.

using that kind of buyer intent insight. But why don't you explain buyer intent as HubSpot as now? Yeah, so it's a little bit different to that. And essentially the way that it works is you help you define an ICP. So you set a series of demographic criteria about the location, revenue, etc. And then you also set a criteria for what you consider intent.

on your website and that's the big difference. HubSpot's buyer intent functionality will not scour thousands of third party websites for an understanding of someone's intent across the web like ZoomInfo does, which, you know, we can't get really around it. It's not as powerful as ZoomInfo from an intent signal basis. However, looked at in isolation, it is a

(16:26.814)
it is a big improvement to the way that the old prospecting tool within HubSpot works. So you define the ICP and then you define the criteria on your website what you consider to be intent. So it may be that three people from an organization visit a specific page a number of times. And if you determine that to be a sign of intent for your website, it might be that you want

at least five people from a specific company to visit your services pages more than two times, then essentially the way that intent works is it combines both of those things and you will be served up records of organizations which meet both of those criteria, i.e. they fit your ICP and they've taken the action that you determined to be a signal of intent. The obvious example everyone uses is a visit to a pricing page, for example.

Yeah, but it might be visits to a pricing page three times and five different people at the organization do it. Because as we know, a student sat in the corner of an office somewhere just scouring prices for products is not intent. But if you have multiple people at a single organization, multiple times visiting a specific page, then you can consider that intent. So the configuration of the way that intent is considered is a massive.

So as we know for those that are more sophisticated in their use of HubSpot, this could kind of been done through a bit of a work around bit of a workflow prior to Breeze being launched. If the data was there, it could pinpoint visits to the pricing page. If the record was in CRM, email, IP address is connected, you could see a potential

prospect is visiting those pricing pages, you could reroute a notification to sales team, et cetera. So what's different now? How has it advanced from that? Well, the difference is HubSpot Intelligence was what came before Breeze Intelligence. ultimately, you could, if you had a prospecting signal, basically, you know, someone had visited your site from a certain criteria, you could add them to your CRM. The difference now is with Breeze Intelligence,

(18:45.036)
when you receive those intent signals, first of all, it's based upon that criteria. And yes, you could do that through multiple different means in HubSpot, but before, now when you add that record to your CRM, it's doing so from Breeze Intelligence, which is all sourced basically from HubSpot's acquisition of Clearbit. So you can essentially get better data from that process. The data that, know, say for example, you saw that

I don't know, Siemens were looking at your website and they did so a certain number of times and you wanted to add that record to your serum. The data that's brought through Breeze Intelligence when you do that is greater quality than it was before. the difference is the UI really, the difference is the way that you can actually configure it, which is much easier, and the quality of the data that comes through Breeze Intelligence compared to what we had before. So that's Breeze Intelligence. Yep. So very timely as we enter a world where

we need cleaner data, particularly if we're to start leveraging AI in any meaningful way. So thumbs up for HubSpot with Breeze Intelligence. So let's touch on some of other aspects of Breeze that were announced to inbound. So where would you want to start with that? What do you think is the most significant next from a B2B perspective? Well, the way that I've been positioning this with clients really recently when I've been taking clients through Breeze,

is what's going to impact you now? What can you implement now that's going to have immediate positive impact? What are those things you should assess in a bit more detail and test, but would have a near term impact? And those things that should really be at this point, you know, experimental in terms of your usage. And that's the way we've been chunking things up. So with the prerequisite of good data to one side,

we can talk a little bit about some of the agents that I touched on earlier. Some of those would fall into the experimental category for me alongside copilot. we'll discuss those things first. So a good example probably with a story from one of those times when I've been taking a client through this is when showing them the power of the content agent. And it is very powerful. don't think we are necessarily questioning that.

(21:12.024)
But I use the example of creating a podcast. Now with the Breeze content agent, you can create a podcast from scratch whereby you have ultimately an AI-voiced podcast, which is a little bit different to this one, obviously. And essentially in a three-minute time frame, I could go from a blog

run that through the content agent and have an audio version of an AI spoken podcast at the other side. I mean, that's the content agent. That's the kind of thing it can deliver for you. The question, I think, when it comes to some of these experimental things with Imbreeze and AI in general, is not necessarily whether it can do that. It's whether you should be doing that. whether you really want to be doing that. Exactly. audience is going to be engaged with something like that.

And the reason for that is that the output is some way off at this point. What you could do, as we're doing now, sitting down and actually talking about it, because you've got, in the podcast example, you've got a fairly good, but still robotic AI voice. You've got no personality there. You've got no visual. You've got question of personality on this Maybe. I've been referred to as robotic number of times.

But, you know, it's very obvious that that is not a real person speaking. think in the world of AI, that desire for genuine thought leadership and personable insight is greater than ever because, you know, it's almost a bit of a vicious circle. AI is increasingly in demand for that because of what it can achieve. And that's just one example. The other agents, social agent, prospecting agent,

are still in private beta. So we've not been able to play with those quite as much. the third, the fourth, sorry, customer agent fits into our kind of planning bucket, our testing bucket. And that's because its usage is more near term in terms of actually being able to see some benefit from it. And that's because essentially what it is, is an AI chatbot, which you can train based upon any assets that you want to train.

(23:38.254)
So you can train your customer agent based upon your knowledge base articles, your website, your blogs, whatever you want to do. You instruct on what it should read and learn and understand. And it can then use that. And by the way, that doesn't just have to be stuff that you own. You can put third party links in there so it can read and understand that too.

Essentially, the way that that works is then it understands all that information and uses that information to respond to queries on a chatbot rather than the pre-configured kind of smart chatbots that came before, whereby you would have to select certain responses and then pre-configure almost workflow style how the chatbot should respond. The customer agent can understand natural language. It can reply in natural language. It doesn't need any configuration for any of those.

kind of branches of activity. that, of all the agents, that's the one I think is the one that will have some more immediate impact. this is what I thought was quite insightful from the conference because Dharmesh, CTO of HubSpot, he obviously very passionate about AI has been, as you said, spending lots of late nights cooking up his ideas. And aside from all the terrible dad jokes in his session. Which he admitted to, by the way.

market on terrible dad jokes, but he's definitely beaten me on that. So Dharmesh's session, he talks about agent.ai. And it was quite insightful, his take on where he sees AI going. And it's very much like you've just said, that there's going to be different agents with different capabilities, different skill sets almost. This whole concept of agent.ai is essentially, it's almost like a professional network for

different agents like LinkedIn. And again, very early days, you can go on there for free and there's a meme AI agent, which doesn't produce the best memes in the world, but it's still a bit of fun to play with. you can see where it's going. It's going to be very much, there's a marketing function and we've got a prospecting agent. We've got a content remix agent. that's what's making up the whole team, combination of AI, human.

(26:01.326)
humans at the core, probably moving to a point where a lot of the hour is creating initial drafts of things. then I saw this from, from, from CMO, Kieran, who talks about how that's the role of the human being that needs to edit that, those initial drafts, try and smarten them up. Cause we all know right now that they're falling some way short of anywhere near quality. so you can see where this is going to hit the nail on the head. And I think.

Our view would be similar that this is a vision for the future, not necessarily a vision for now. And that's why they're in that experimental bucket for us, because it's very clear that whilst they're not necessarily going to revolutionize your marketing function overnight, as a result of the announcements that have been made with Breeze, they are the direction of travel for how you're going to use HubSpot going forward and probably how you're going to use Most Tech going forward. And I suppose that brings us nicely onto Copilot because

We're all used now to using ChatGPT and Claude and that kind of prompt-based work that most of us are doing these days. And that's coming into platforms like HubSpot. So many of the agents and copilot, which we'll talk about in a second, are prompt-based rather than ultimately doing the do. I'm sorry, I don't want that in the podcast. Doing the do? Yeah, they're prompt-based.

They're prompt based rather than you doing all of the work directly yourself, essentially. And that is the future for the way that you will use HubSpot going forward. But it is not the future. It is not the now. But that being said, to prepare for that future, we're recommending to our clients that they experiment with these agents. They experiment with the copilot functionality, which is essentially chat GPT within HubSpot.

The important difference with Copilot is that it's reading your data, unlike ChatGPT, which is obviously reading the internet, basically. Copilot is informed by the data within your CRM. So in theory, it should produce much more reliable, much more accurate, more unique, much more quality Depending on what you're looking for. I think, you know, again, Copilot isn't the same as GPT. No.

(28:27.564)
HubSpot Copilot. It's not the same as Claude, particularly the more advanced pro versions, which we use and are great. When you search for inspiration, get started on a task or what, you know, they really do fill a gap and help me to work a lot quicker. But that's for general tasks. This is very much specific to what you need in HubSpot. Yeah, I can give you good example of that. So anyone that's used HubSpot will have spent hours and hours and hours building complex workflows. That's just, you know, most people.

And sometimes it might be quite a simple workflow, but this just takes a lot of time to put together. Now, Copilot can, and it can do now, it can produce a workflow based upon a prompt. So again, I was demoing this to a client the other day and I asked Copilot to write me a workflow that essentially rooted leads for a particular region from a particular form to a specific individual. And it did that. It did that fine. It worked. It did it.

but it was a simple workflow and it wouldn't necessarily save you that much time doing it prompt based versus actually building the workflow yourself. Now, where you would save a lot of time is those long complex, complex, long complex, difficult to put together workflows. And some of the initial tests I've done trying to build similar things, it's just, shit, it's pan, to be honest, it's fallen over.

And to what is a better phrase? you know, ultimately asked me to re rework the prompts and things like that. So there's, you know, there's a way to go there, but the division for how it could work going forward once, you know, it becomes stronger and it becomes a more powerful platform, you know, is really, really clear. And, you know, that workflow one is a good example. And I think it's, you know, some people could be really negative about it, but for me, it's like, well, look, there's going to be a lot of change. We're going in this direction now.

and there's no doubt that that's the direction of travel. So we can't really fight it. Yes, the technology is still quite early stage. it does give us an opportunity to start to change the way we think, change the way we work with this technology. Whilst it is a very sort of basic level, as it does get more advanced, at least we're into the way of thinking, we know how it works. I think just right now it's about experimentation.

(30:50.99)
relatively low expectations in terms of what... you try not to be dismissive. It's not going to produce a fantastic brochure or an ebook or, without you, without any real intervention. I wouldn't want it to necessarily create a podcast because I certainly wouldn't be interested in watching a couple of robots speak on a podcast. It might be better than this one. Probably better than this one. But I still think that it's very timely when it's come out, when it has, from the data perspective, but also...

gives people now who are quite nervous about AI and what it means a real opportunity to just experiment, touch, try different things out as the platform evolves. Naturally, everything will switch to prompt based control of your platform. Now is a good time to get your head around that and start experimenting. And like you said, don't have too high expectations. It's so easy to be dismissive. Yeah. Now, the first time that copilot

comes back and says, sorry, I can't help you with that. It's easy to get out of that shit. I'm not playing with that anymore, but you know, ultimately- could leave it alone for six months and come back to it. So, whoa. And you can get overtaken. You can get overtaken in that time. Yeah. So the way I'm using HubSpot these days is co-pilot first. Yeah. Now I'm ultimately thinking I have a task in mind that needs to be completed. I'm asking co-pilot and you know-

half the time, but maybe even more in certain circumstances, it can complete that task for me, or at least get it started, give us a starting point to work from. Some of the time it's not, but then I can go away and do that work and I've still saved. The best example for me is content remix, because you can take a blog or something as a source piece of content, run it through remix, and let's face it, what comes out is pretty shit. I'm particularly on that. That's not the point. The point is...

It's demonstrating the capability of the tool. And it's showing the art of the possible. as long as you've got the right expectations and don't dismiss it at hand, because obviously over time, as you train these tools to, you and you input the right information, the right background. And you train yourself on how to it, it will start to become more and more effective. So you'd be a little stupid to start dismissing it based on that initial experience. I was a little worried when

(33:15.67)
I HubSpot was very bullish as always when they announced these things. So I was mostly interested in what the investors would think of the new announcement. I paid a lot of attention to the investor session that they do on the first day. Again, HubSpot very bullish about the numbers and know, they've grown another 20 % this year. And it was interesting before the investor day, HubSpot had a whole rating on it from the investor community.

And it switched after the conference to a buy rating. a lot of what was said at the conference in the analyst session was well received. A lot of it was around, they go to market model, they're sort of high end taking on Salesforce and the volume play, mid market trying to get new customers on board using the starter licenses. Very impressed with the CT model that's changed recently. But also, I think,

expectations were kind set at the same level we've just talked about around AI, that this is emerging technology now built into the platform and it's just going to evolve over time. And touching on Remix, think if you can look at Remix as, and also Video Remix was announced, and that could be, we haven't had a chance to play with that one yet, might be able to with this podcast, but that one is something which I think could be certainly within that testing phase rather than experimental,

Essentially what that does is chunk up a wider video and you can create snippets from it quite quickly. Video editing is, as Emma will tell you, on our team, a very laborious task. takes a lot of time. And so that's one definitely we'd have on our kind of have a go at now and it could have a big impact straight away. But content remix, think, if you look at it as a starting point, if you look at it as an option that you can edit, and by the way,

images completely ignore because they are absolutely awful and sometimes quite scary. That's the case for all AI image generation at the minute almost. But if you look at what remix spits out as a starting point to either trigger some thought on your end and think, I don't like that. This is what I want. Skip that kind of writer's block that you get. Then that's a perfect use for it. I think also landing page builders. So for example, you can put a blog in, it can build a landing page about the same topic.

(35:41.174)
If you get your template well, if it's built well, then it can do a good job of that. again, it's a starting point. If you're going into it thinking, okay, I'll put this in and what comes out, I can publish and move on with my day. Away from that. But if you go in and think, okay, well, it might prompt some thought at the very least, and maybe it'll go a little bit further and I can just edit what's there. You know, might have to do quite a lot of editing, but it starts you off on that path at the very least. Yeah. And I think it just raises, if you have that iterative approach,

you'll get there a little bit quicker maybe than without the technology and certainly probably at a higher quality as well because you're getting input from another source that's aggregating information from elsewhere that's just going to fuel your own creativity. I think we probably should sum up. And I think for me, main takeaways from, I'm not going to talk about inbound, everyone talks about inbound, but the features and functionality that's come out. Whilst initially I was very excited by, because I was caught up in the whole kind of

conference and the rah rah that HubSpot are so very, good at. Now I've had some time to digest it. Went through that kind of little bit skeptical almost to this point where actually when you realize where they're at and what they're trying to achieve, it is really exciting. It's it's very early days still. as always what HubSpot does brilliantly is it makes things simple. I think Salesforce,

Maybe they run their conference a week earlier or similar time. Again, very much focused on AI. I dread to think what AI would like working in marketing cloud and Salesforce. I can't imagine it's anywhere near this sort of elegant experience that you're to get working in HubSpot because it is designed for the users. So that would be my key takeaway. Sure about you? Mine would probably be more practical. I'd say analyze all the stuff that was announced and it's part of Breeze.

Certainly some of the AI stuff, start to experiment with, have a look at things like form shortening, test them, build them into your business if there's a good use case for it, but don't just do it because it's there and it's new. If it's not going to provide you any benefit, then leave it. And then there are certain things which I would recommend that everyone should start to look at implementing now. That's lead scoring, which was announced at the same time, but isn't necessarily connected to Breeze.

(38:06.03)
which is a massive improvement on the old bead scoring model in HubSpot, and that's in the Intent Signalster. Those two things, I think, start with work through the rest and experiment with the AI stuff. And one question. Out of all the companies that you could have referenced earlier, why did Siemens spring to mind first? No comment. I think that says a lot about you, James. I don't know. I think we can call it a day at that point.