INBOUND24 Rebound Session Breeze

I am going to cover up the chat with my notes. So if I'm not seeing something, Kyle, just yell at me.

Yeah, we'll call it out.

Also, people have questions, just put them in the chat. I'll like stay online after my session. I'll like respond to stuff individually too. So I was told I only have seven minutes, so I'm going to dive right in. I'm Adam. I'm the product manager for Breeze Copilot and I'm going to show you a couple of different use cases from a sale, sales, marketing and service perspective. So I'm going to start off with sales. So I'm here inside of HubSpot. As probably all of you know by now, there's lots of different ways that you can get into Copilot. The easiest way is just to invoke Copilot in the top right hand corner here. And the thing I love to do whenever let me actually start a new chat here for any new users that is new to Copilot. You've never used an AI assistant before? What I always recommend is check out this episode see more prompts button. And this is a great way just to get a summary of all of the prompts that you have available to you in Copilot. So if you're not really sure what types of questions that you can ask, this is the place I recommend going. It just gives you a really great template for different types of queries that you can send to Copilot. Let me actually expand this out since I'm on a pretty big screen here. So I'll make Copilot full screen. And the first use case I want to do is let's go ahead and do some research. So we're going to imagine we're a bdr. We need to do some research on a prospect before we can add them to to our CRM. So I'm going to look for one of the research prompts and I'm going to do research a company and you can see when I actually select one of those prompts, it gives me this nice little Mad Libs style. So again, if you're not sure how to type to Copilot yet, definitely use these prompt suggestions. It just gives you the format for you and then you just have to fill in the blank. So I'm going to say research this company. Loom.com I'll hit enter. Copilot's going to go off and do a little bit of thinking and what it's actually doing is it's quoting, querying Some of our proprietary data sources that we have. So if I was to do this in ChatGPT or Bing or Gemini, basically what I'd be doing is just a Google search. But what we're actually doing here in Copilot is we're plugging into Breeze Intelligence, which is the artist formerly known as Clearbit, and we're getting a whole bunch of extra data about this company from those data sources as well. So I'm going to get things like funding, information company technology, stack, industry information and things like that. This is taking a little bit longer than it usually should. I may need to kick and restart this again. Let me try that again. Let me do that again. Research the company. You can see I can just type to it as well. So research the company. Loom.com. let's give it a second again. There we go again. A nice little summary of Loom here. And again, I'll expand this out. When I expand it out, I get a lot more details here. If I open up the context, I. I can see lots of information about this company, their industry, the tech that they use, social media, links, number of employees. I still need a little bit more information here. I need to know a little bit more about their funding. So I'm going to ask. Tell me about their funding just to make sure that they have the budget for us to support them. So Copilot's again going to go off and think and then I get some information back about the different rounds of funding they've had all the way from C. Proceed up to series C here and perfect. This is more than enough information for me to know that I can add them to my CRM. So I'm going to actually scroll back up and I'm going to add loom to my HubSpot and then copilot is going to go off and create a record. And I can do this for any CRM record type. So if I have contacts I need to create, I have tickets or deals I need to add to my system. I just need to provide Copilot with an email address or a name for that deal or the deal amount. And it's going to be able to create that record for me inside of HubSpot so I can verify that that was created. Sorry, I realized you can only see one tab. So let me just navigate manually there so I can see if my will load. There we go. You can see I have now this company has been added and if I was to open Copilot Backup again, notice that it just continues off exactly where I left off in the conversation so I don't have to re prompt Copilot again or go back to an old chat. Anywhere I navigate throughout the system, it's always going to be there for me. Let's say I now want to assign this company to myself. I can say assign them to me. Notice that I didn't even have to specify who I'm talking about. Copilot has a bunch of context, not just on the conversation, but also the page that you're looking at and knows who me is. So it's going to be able to go off and assign Loom to me. So Loom has been successful, successfully assigned to you in HubSpot. If I was to refresh this page, I'm going to see that they have indeed been assigned to me. Okay, what else do I want to do? So let's say I want to. Let's kind of move up a little bit in the kind of sales hierarchy. Let's say that we're now a sales manager and I just want to get a summary of a bunch of deals that I've created in my pipeline. So I'm going to ask Copilot here, hey, can you show me a list of all the deals that have closed in the last, let's say, 12 months? Copilot is going to go off and it's going to plug into our report builder and it's going to generate a report for me. So it gives me a nice oops. Looks like we have a pretty small data set in this testing system here. Sorry about that. So it gives me a data point here. Let me actually do a slightly different query. I'm just going to say show me all contacts that have been added to my CRM in the last 12 months. Make it a line graph. So I just want to show you kind of what it looks like with a little bit more data. So I'm going to ask it to get me a list of all of the contacts that were added in the last 12 months. And I've asked it to make it a line graph. So Copilot's going to go off again. It's going to plug into our report builder and it's going to give me a nice little line chart for that data. I can edit this in Report Builder from here. And again, I'm not sharing my whole screen, so I'm going to just copy the tab that opened for me there, paste it in. Then it took me into the report builder. Whoops, sorry, wrong one. Let me do it again. Hold on. I'm going to actually cancel my screen share. I'm going to do this once more. Sorry, bear with me. I'm going to share a whole window instead. Just so you can see everything I'm seeing. Share. Okay. There we go. Okay. So this is what opened for me. So it opened up the report after I built in Copilot. And this is great because like, if I'm like, let's say I'm needing to build a report for my board or something like that. Often the hardest part of building a report is just getting the initial report built out. So I can go to Copilot for that, give it a really quick query, ask for a line graph. I could also ask it like, hey, can you give this to me in a pie chart? Or something like that. It will build a report for me and then I dive right into Report Builder. I can add additional columns, I can save the report, I can export it if I need to and share that around with whoever I need to share that report with. So Copilot is a great way just to get started with some of those more complex reports that you need to build. So that is Report Builder. I'm going to switch gears now and we're going to take a look at a. Let's take a look at some marketing use cases. We're going to take a look at writing a really basic blog and doing some refinement to a blog. And I'll show you some of the refinement features that we have in Copilot for editing content. So I'm going to start a brand new chat and I'm going to imagine here we're a company that sells Houseplants. So I want to write a blog post about, about how to take care of houseplants. I'm going to say, can you write me a blog about how best to care for houseplants? Copilot is going to go off, it's going to invoke its text generation command and it's going to write me a really nice blog here. I do have some refinements that I want to do to this, but I'm going to go ahead, let it finish writing it first and then I'm going to create it a blog draft inside of HubSpot. So again, we can see that it has created a full blog draft for me. And then what I can do from here, let's say that I want to make some refinements to this first paragraph. I think this is maybe a little bit too long. I can just double click on Highlight that entire paragraph. I'll see this new button appear here for refining this paragraph. I'll click on that and I'm able to rewrite, expand, shorten. I can change the tone if I want to, or. Or I can just open up copilot itself and give it a completely custom command. So I could say, hey, can you rewrite this in the first person? Maybe I want to make it a little bit more personal. That's how I like to start my blogs. And so I'm going to say rewrite in the first person. It gives me a preview of the text that it's rewritten and what I really like about this compared to some of the copying and pasting that I did previously in chatgpt. Here I can see the new text on the right and the current text here in the actual blog itself. If I like what I see, I can go ahead and hit replace, and that's going to replace that text right there. I have just a couple of minutes left, so I'm going to finish off with a real quick use case. Now, for service, we're going to look at a couple of things for service. I want to just show you some of the basic summarization tools that we have. So if I go up here to search, let's say that I have to get up to speed really quick on a customer that I'm going to be talking to today. I know her name is Becky Hall. So I'm going to search for Becky. Now, inside of search, when I hover over the name of one of my contacts, I'm going to see this new summarize button, and I want you to memorize this button. You're going to see these copilot buttons showing up everywhere throughout HubSpot. And these are just your little indicator to let you know that, hey, you can ask copilot a question about this particular record. So here for Becky, I can just click on summarize. I'm going to get a quick summarization of Becky's account and I can just regenerate there just to show you how fast it is. So I'll hit summarize and then I get a nice summary. Becky hall is a lead she's created in our CRM on September 6th, I can see what company she's associated with. Oh, I see there's an open ticket. I probably want to take a look at that. I can see that there's a phone number, email address, and some tasks on her record. Let's take a look at that ticket. So I'M going to ask a question that's going to pull open copilot and I'm going to say, can you tell me about the ticket? And notice again, I don't have to tell copilot who I'm talking about. I don't have to specify which ticket I'm talking about. Copilot has all the context of the conversation and it's able to let me know. The ticket title shipment issue was created on September 10th. It's currently in a new status. The tickets owned by Adam Stevenson. And I can click in here, I can see more details. I can open that up and I can drill down further if I want to. Okay, let me finish off here. I'm going to show you some features in calling. So I'm going to pop into my CRM. I'm going to go down to calls and let's take a listen into one of these phone calls that we have here. So I'm going to click into this first one and I'm obviously not going to play the full phone call for you here. But you can see we have a five minute call. It was a long issue in this particular phone call. Again, maybe I'm just getting up to speed. One of my colleagues had taken this call initially. I don't have time to listen to all of this. So instead, what I'm going to do if I close this up, what I'll see when I first come into this call again, I'll see this copilot summarize button. So remember, look out for these star buttons. These are your indicator that, hey, there's a little bit of breeze AI here that I can take advantage of. Somebody say summarize this call. Copilot is going to go off, listen to that phone call, will transcribe the transcript of that call and then I can get a nice summary. So I have Lucy from this organization and Adam spoke with Adam to express frustration over a delayed planned order. Adam found the issue canceled. It offered a discount, expedited delivery at no charge. A couple of more details here. Lucy. It sounds like she asked for a follow up. So I'm going to go ahead and ask a question here and I'm going to say, hey, can you help me write a summer follow up email to Lucy to resolve the issue? All right, again, copilot is going to go off. It's going to write some text for me here and it's going to generate an email. It's customized it to Lucy. It's put in some specifics from that ticket. I didn't have to specify what issue. Again, it knew from the context of the conversation that we were talking about this. This phone call that we had so included some details about the discount that was offered. Put a nice little sign off here. And then of course it included my information. From here I can copy this into my email editor. In the future we are going to be adding features where you're able to directly take emails written in Copilot and insert them into emails that writing either in your CRM or inside of Gmail or Outlook, which with the HubSpot browser extension. All right, that's all I have time for. I could have done a whole hour on this, but I'll stick around for a little bit just to take some questions here.

Amazing. There are a lot of questions in the chat if you want to dig in. I also see the Q and A number ticking up. So look around. There's plenty to respond to. But in the meantime, let's pass it over to Amanda.

Hello everybody, My name is Amanda Schumacher and I am a product manager over in Content Hub. Just give me one moment to get my screen share set up. I'm going to share my entire screen and I'm going to just go and make sure that we've shared. All right, so hopefully everybody can see. So my team built the content agent jobs to generate blogs and case studies. So for our demo today, I'm going to show you a high level overview of the content agent, sort of homepage and settings and then I'll give you a demo of the job of the content agent to generate case studies. And there are four total content agent jobs. You'll hear from my colleagues Lucas and Kelly in a little bit and they're going to give you a more in depth demo of generating podcasts and landing pages. So just to give you a little bit of context before we dive into the ui, I'm going first and I'll be the first to talk about the content agent. So I'll set the stage. Content Agent is available at Content Hub, professional and enterprise. And this content agent that you see is currently in public beta. So you'll need to opt in to get access. And in order to use the content agent jobs, you'll need to have permissions for the respective content types that you are generating. So if you want to generate a blog post, for example, you'll need write permissions for the blog. Now, why did we build this thing? Producing a steady stream of content can be daunting for companies, especially companies with small marketing departments or limited Budgets. So the Content Agent is like an honorary member of your marketing team that can create content drafts for you as a starting point, so you'll have more time to focus on high value tasks like your strategy work or creative pursuits. It's like having an entire content team at your disposal, but without the overhead. So now I'll take you into the product. We'll begin our journey in the Content Agent Breeze sort of home space. The Content Agent home space you can find Linked. From here I'll click Launch and it will take me to this home page. So this is a single central location where you can access all four of the Content Agent jobs to create landing pages, podcasts, case studies or blog posts. Now, before you get started using the Content Agent, what you'll want to do is you'll want to make sure that you set up your Breeze AI settings and data sources. And that is because the more data you provide Breeze about your company, the more accurate and personalized to your business the content is going to be. So you can access your AI settings either directly through the Breeze homepage or from Global settings. Under AI, there's going to be this access tab. This dictates who gets access to the tools. In order to access the Content Agent, you'll need generative AI access as well as CRM data for case studies and then you'll go over into this data sources tab here. So a few things that I will highlight. You can set up your ideal customer profiles. These are basically your target Personas and why it is valuable to set these up is that we will allow you for many of these content creation jobs to select from multiple Personas if you have them. And we'll try to tailor the content more specifically towards your audience. We also allow you to set up a brand voice that you can use when generating the content. So in this area you can set up a brand voice where you can train the tool on different samples of your content and give different personality descriptors. And once you set this up, any of the content generated by the content Agent will be in your unique brand voice. So I show you this first because this is one of the places where Content Agent really shines over competitors like ChatGPT. It's intelligent and it intimately knows your business based on your Breeze data sources. And as you'll see in case studies, it can also connect to your CRM. It knows your brand voice and for some of the jobs it can even look at your other content. So now that our sources are all set up, let's take a spin through the case study generation job. In this example, I am a fictional company that is a plant daycare named Zen Plant Care and I want to create a case study about how I helped my customer Data Hound transform their workspace through greenery. So I am going to just start off by entering in a prompt. This is where you'll give context to the AI about the case study that you want to create. So I'm just going to write case study about how Zen Plant Care helped Datahound. From here you'll be able to select the country of your target audience, which we use for SEO keyword targeting, and you'll also be able to select a company. Now this is what I was talking about when I said that the case study generation job is directly connected to your CRM. Here you'll see a list of all of the companies that are in your CRM and when you select one, we will use this information to help pre populate part of the case study. The case study, for example, has a section about the company and so we'll look at the properties on that company record and attempt to fill out parts of it for you to save you valuable time. You also have options to choose your Persona, which I showed you in the BRI settings as well as the language. So on the next step I will have the option to generate an SEO optimized title. It will give me some suggestions for the title of my case study. All of these are editable and wherever available will recommend SEO keywords as well. We have a partnership with Semrush and so we are able to pull keyword data for the number of Google searches for a given keyword in a month, how hard it is to rank on Google for that keyword, and then the goals of the audience when they're searching for that keyword. So are they just looking to find a webpage? Are they looking to buy something? Are they looking for information? All things to help you optimize your case study, depending on your goal. And on this final step, this is where a lot of the magic happens. So let's say for your case study you did a zoom interview with the client. In this case, Datahound, if you have a video file of that zoom recording, if you have an audio file, a transcript, this is where you can upload all of that context and the case study agent will review all of those notes, try to pull out any relevant information like metrics, quotes and testimonials, and attempt to fill in the case study for you. So let's see what I have available to me here. For this example, maybe I have a call transcript. I'll go ahead and click Generate and then the case study will process and generate. Just for the sake of time, I'm going to flip over to this tab and show you one that's already generated so that you don't have to sit here and wait. But you'll also. You'll see here that we have a brand new standalone case studies app with a new template and we will try to generate a summary of the challenge and results of the case study based on the files that you've uploaded. We'll attempt to pull out and populate any metrics that are mentioned in your files. We'll even be able to fill out the quotes and testimonials that we find in your file, as well as this section which is populated based on the CRM company record. All of this is a WYSIWYG editor section, so you can change anything that is not accurate or put your own personal spin on things. This is really just meant to be a running start to help you get to your end case study more quickly. So with that, I've shown you the case study creation flow and I will open up for any Q and A from the audience.

Great. Yeah, we'll also have you handle that asynchronously. The Q and A number is ticking up. You should be able to respond to some of those by chat. But let's go ahead and move on to Andy just for the sake of making sure everyone gets a turn.

Hey everyone, My name is Andy Lozano. I am a design lead here at HubSpot, working on the prospecting agent. Going to show you what we did and what we made. Okay, so this is what you'll see when you have set up the agent. Let me really quickly just go through what onboarding looks like so that you can make sure you're giving your agent the right context about you and what you sell. This is what you'll see. If you've never set it up before, go to set up agent and then the agent will automatically pull in your portal domain. But if there's a different website that you want it to know, you can change that in here. So when I go to next, it's going to scrape that domain for all of this information for me. You could obviously edit make changes in here if you didn't like what it said. It's going to show you the company, what your company sells, your value proposition, what pain points your company solves, also what industries you typically sell. Into, you can remove or add more in here, as well as what Personas or job titles you target. This means that whenever you go to enroll somebody in the agent, it will give you a notice like a badge, a tag that tells you this is one of your target Personas or this is one of the industries that you target, and all that in a little bit. Okay. Next thing you want to do is select who the user acts on behalf of. So that could be a member of your sales team, or if you don't have a sales team, it could be yourself. You could connect a different inbox in here. It's going to pull in the emails that you have connected to HubSpot today. Then you can also tell it what meeting link you want it to book on behalf of. So this is how the agent's going to help you get meetings for prospecting agent. And then the last step is you can select the level of autonomy that you're comfortable with. So we recommend starting with semi autonomous, which means when you enroll someone to the prospecting agent, it will research the contacts that you selected, it'll write you three outreach emails, and then it will ask you to review the emails before they're sent. So starting here is a good thing because then you can make sure you like what the agent is writing, that it has the correct selling information and you like how it's going. And then fully autonomous is maybe when you feel a bit more comfortable and you trust what the AI is saying. So that's going to research, write out reach emails, and then emailing those selected contacts for you without asking for review. So I'm going to stick in semi autonomous mode. Once I create that agent and I've started to use it, I'll see that there are two ways that I can enroll people to the prospecting agent. There's automated enrollment and manual enrollment. So let's jump into automated enrollment really quick. This is a new thing in HubSpot that's more like contextual automation or contextual workflows. They're like workflows light. I'm sure many people on this call are very familiar with the workflows tool and how it works. But one thing that we learned from customers is that it can be quite overwhelming when you just want to set up a quick automation. So this allows you to do that. Basically you come in and say when forms are submitted, whichever form you want. In here I could say also on the registration form. That means when someone submits that form, they will be enrolled to the prospecting agent. So I'LL save that days since last engagement is a cool one. This will help you make sure that no lead goes trolled and someone's always warming them up. So in this case someone will be automatically enrolled in the prospecting agent when there's been 30 days of knowing engagement. And this is like forward looking. So from today it will say the 30 days. So it's not going to do a backlog and enroll thousands of contacts into your agent. And then we also have page views. So like I said, this is a new thing In HubSpot. We are an active collection of these use cases that from our customers and from our private beta participants to make sure that we can get quick contextual automation for you three that we started with. And then we also have manual enrollment. So let's say you've got somebody at inbound and you want to enroll them to the prospecting agent. I can come up here and I'll manually enroll contacts. I'll come in and say oh yeah, I met Kendall Royce from Buffalo Trace at inbound. Thor Odinson and let's do Beth Dutton. All right, so I've got those three contacts selected. Oh, I have too many in here. So we won't do Beth, but let's go forward to next. And this is kind of like your review your cart before checkout. So all those things that we set up on onboarding, this is just to confirm that you like what the agent's going to do. It's going to stand on behalf of me, it's going to book these meetings, it's going to write three emails in a witty tone and I have semi auto mode on and it's going to stop the outreach when I get a reply or booked meeting. So I'll start enrolling. And now what happens is that we see the agent is currently researching the contacts that I assigned. But let's say I've already assigned folks and I want to see what the agent said so I could go in to review the emails for this particular contact and see that the agent wrote these three emails. I'm going to review them and then when I feel comfortable with them, I'll click Start outreach. I can edit these emails right in here, make any changes that I'd like. I can see that there's going to be a series of three mails sent right in here. So when everything looks good to me, I'll click that. Start outreach. And now Jane is being handled by the prospecting agent and the prospecting agent is queuing up those emails for me. So now I don't have to worry about it or try to remember when to send these emails to Jane. Excellent. We got an error. You know, live demo is happening. Our engineers fixing some things. All right, so let's see what happens when the agent is done. So in a finished state, this could be like all of the Somebody manually enrolled someone from the agent or they've finished because somebody has booked a meeting. So it would look like this looks like I sent one email to Jane on September 18th and Jane booked a meeting with me shortly after that. So after one email, I got an email or a meeting from Jane. That's what the finish date looks like. And then a couple of other things when we went through the setup experience, the first time experience, you can always come back into this agent setup and change. So if you don't want the agent to act on behalf of you anymore, another user could come in and kind of claim the inbox. Right now, the agent acts on behalf of one person, so you only get one email address for the agent. I'll be honest and let you know that that's been pretty limiting from what we've heard of our private beta participants and what we heard from people like you at Inbound. But we're actively looking at how to open up that functionality and to give more inboxes to the agent. There's all that selling information that we went over before. This is the outreach guidance. So it's just saying that there's going to be three emails sent to the enrolled contacts in this witty tone. Personally, I like the way the WIDI emails are the most. I think they're the most fun and they sound the most human. You could also set different lengths. Right now we have the default as 4 based on some user feedback that we have. But you can change it in here and then each email is going to get progressively shorter. Also, the length is set by the prospecting agent for you based off of what you set up here. One of the cool things about the agent is that using its research, it will come in and write these emails, but then on the second email it'll try a different angle from its initial research. So it's never going to try to target the same pain point again or the same value proposition. It will try something new and how you can see what that evidence is. If you go to any of your enrollments right in here, let's click on this other Jane. There's this View Evidence panel that will show you what it found about contact and company research from your CRM. If there was more that the agent found. Let's see if we can find one in here. Yeah, this one has quite a bit. There's CRM research from both the contact and the company. There's a bunch of web research. So what it found out from Daisy Works at Dropbox. What do we know about Dropbox? Also, recent company news is a really cool one as well. You could see if the agent did end up referring to a recent article or recent news that was dropped. You can see why the agent referred to that way and double check if you'd like. Okay, a couple of really quick things. I'm sure admins on this call might be wondering how did they give or take away access to some of their users? So if you come into user preferences, I'm going to go to user and teams. I can give or take away access to any user. If you're a super admin, you automatically get access to the prospecting agent. But coming into here, go ahead and say, wow, we are really slow. Where's my edit permissions? Coming up. There we go. We'll just do Josh. I can come into edit permissions. I've got sales. Oh, he's a super admin. Never mind, we'll go to Dylan. All right, I'm going to say that Dylan does not have access to the prospecting agent, but I'm going to go ahead and grant him access. Access can turn that on and save. And now Dylan can see the prospecting agent in the navigation. So you can always take away any access or give more. It's up to you. As the admin last thing here kind of touched on AI settings just a second ago, but these are the global AI settings that need to be turned on. You want to give access to users to use generative AI tools and make sure CRM data and customer conversation data are turned on and then the agent can work its magic for you. All righty. That's all I got. I will definitely hit. I see the chat is going up. I'll hit any questions on there and hand it back over to you, Kyle.

Amazing, Jess. Moment of truth. Let's see if your screen will share.

Fingers crossed. Did that work? Okay, what do y' all see?

All right.

Perfect.

Alrighty. Cool. Well, hi everyone. My name is Jess. I am the product manager for the customer agent, previously known as the AI chatbot. And today I'm really excited to walk you through what it's going to look like to set it up to engage with your website visitors. But before we get started, let's just talk a little bit about like what the customer agent is and the value of it. So for starters, it is a GPT powered tool. It's going to be able to respond to visitor questions using HubSpot content you've already created. So if you have a knowledge base you have hosted on HubSpot blog posts, it's all going to be able to answer questions using those materials. It's also going to be able to scrape public URLs so maybe you don't have a ton of content. It will be able to grab text off of your public website or other URLs to be able to generate content for your visitors. So in minutes it's going to become your most efficient team member. So if you're on the breeze page and you scroll down, customer agent will be at the bottom and here if you go to launch, that's where you'll see where, where you can access the customer agent. So in short, it's going to become your most efficient team member. It's going to free up instantly resolve tickets and free up your team to handle more of the high value work. And it will also seamlessly hand off conversations to human agents as well and ensuring that the handoff is really smooth and that your customers ultimately get the answers that they need. So let's talk about. We're going to use the example of Chris for this demo. Chris is a customer support manager at Bike Tools. Their company is growing rapidly and so this means that there's a lot of website visitors and support requests and it's great for business, but it puts a lot of pressure on his team to deliver those fast personalized responses with fewer resources. Chris sees that HubSpot has launched the customer agent and he's excited to get started. He starts by hitting that create agent CTA and the first step is to actually create the customer agent. He's asked to give it a name, a role and a personality. We're going to go with Bob here. There's three different options for the role. Customer support, marketing specialist and sales representative. Most of his incoming volume is for like support related questions, so he's going to go with that role. One thing to know is that right now you only have Chris can only create like one customer agent for his portal. But in the future it's something that we're looking to expand to. So have multiple agents. You may have like multiple agents with different roles and personalities next. So we have the personality. Well, let's try witty. I think Andy was saying that she loves the witty responses, but yeah, let's. I think that's what Chris wants to go with. So let's try that. And then here the next step is to actually give Bob the content he needs to be successful. So I talked about how you can train it on your existing HubSpot content. Chris is going to already has Knowledge Base articles, he has a website and some blog posts. So he's going to go with that. It looks like he can always add more content later. So he's going to hold off on adding the public URLs for now. So that's it. He creates the agent and just like that, Bob has been created. This is the overview page. Chris gets to this page and sees two things that he can do. The first is that he can actually start testing Bob and see how he is going to perform and respond to visitor queries. So I'm going to ask. I had a blog post I think about cleaning bike chains. So I'm going to ask about that. How do I clean my bike chains? Can you help? That's maybe. Let's try a different one. Maybe it wasn't bike chains. Hi, I need my bike repaired. So Bob is thinking response. The two wheeled Steve needs some tlc. Are we talking a squeaky chain? A flat tire? Does it need the full spa treatment? Let me know so I can steer you in the right direction. Definitely. Pretty witty. And let's try it. So that's great to see. You know, Chris sees that it can respond and let's try asking another question. And I, you know, let's go back actually to the content page. I don't remember exactly what I was, what I trained Bob, what Chris trained Bob on. Let's see, this is all around. Let's just copy and paste this question right here. I think there are some things happening because this usually should work. Again, problems with the live demo. But typically what would happen is that Bob is going to be able to respond and actually generate a source citation. Let me see if I can try another question. All right, we're just gonna try the visitor experience and see if that makes it any better. But here, this is a step where again Chris can see if Bob can answer questions. If not, he can actually go to the content tab and add more content and that's where it'll be able to generate those personalized responses. The second step that he sees on this page is to actually assign the agent to a channel because if it's not assigned to a live chat, it's not going to actually respond. So let's assign Bob To I have two chat flows here, live chats. I'm going to sign it to this first one which is routed to the help desk and it looks like it's already on. I'm going to set it to, you know, in the event that Bob can't respond, I'm going to have it assigned to me. And there we go. All right, so now let's go to the visitor experience and hopefully this is a little bit better. All right. All right. We see that Bob is able to start by Jimmer, you know, asks what I what he can help with. I need help with my bike. It needs to be repaired. Fingers crossed. I think there's something happening on our back end. Not sure why this is not working, but here I can go back to like previous conversations and I can show you how do I start trial of any bike tool, bike tools tools. And here we have a really thorough response. And earlier I talked about source citations and it'll actually take you to the source in which it used to pull the response from. And that builds a lot of trust for that end user, but also for the admin to know that this is something that wasn't hallucinated. It was a response that actually is backed by honest material. Chris is feeling really great overall about not obviously in this demo, but in general he feels good about the sources he's trained Bob on and he can actually see how Bob performs directly within the help desk. So these are all the tickets within the help desk. This is, you know, the conversation that we just had around not actually knowing the answer. So at any point in time I can actually step in here, step in to actually take over the conversation or you can see, see in previous conversations too how it was able to resolve. Let's take a look at this one over here. This one was just bananas, but maybe the one before, you can just start to see all the conversations that the, that Bob is having. So overall, let's now go over to back to the app and we're going to check out the performance tab. This is where Chris can see all of the metrics that are being generated for Bob and you'll be able to see things like how many conversations were resolved, how many were deflected, how many were handed off, how many times it didn't actually answer questions, and the overall visitor feedback. There'll be like a pie chart that shows you if it was helpful like the visitor said it was helpful or not. But yeah, this last tab is just all those initial settings. So if I wanted to, Chris wanted to actually change it back to friendly. He can do so. I'm a little bummed that the backend is definitely down right now, but have any questions or concerns just about the response quality, let me know and I'm happy to connect but that's it for me.

You're getting lots of love in the chat, Jess. It's a very sympathetic crowd. I think everyone here has experienced experience something similar to this. Still, it's a cool feature and I appreciate you coming here sharing what it's supposed to do. We are doing great. We are on a sprint to the end and I think everyone's going to get a turn. I'm really impressed with all of you and your time management skills. Let's go to Kelly next and then Lucas, you'll close us out.

Hi everyone, my name is Kelly Yuen. Pronoun she her and I will be demonstrating how to use Breeze Content Agent to generate landing pages. So let me go ahead and share my screen. Perfect. Can everyone see it? Okay, so I'm going to be demonstrating how to start generating landing pages and what else you can expect from the flow. As Amanda demonstrated earlier, you can access the Content Agent from the Breeze overview page. You can launch the Content Agent and then you can go ahead and launch the Landing page job. Specifically, you can also access the Landing page job from the Global Nav. If you go to Landing Pages and then you can click Start with AI. So I'm going to go ahead and click Start with AI and on the first step of the landing page flow, you'll be asked to identify the purpose of the landing page. This helps us understand and helps us select which template we'll use when generating the page. For example, if you select Get Visitors to book a meeting, we will add in a default meeting module to the page. If you select Run an Event and Collect signups, we'll add in a default form. We also want to highlight that if you select Offer Downloadable Content, you have the option to upload a file that you're offering. We'll pre create a thank you page so that once your visitors fill out the form, they'll be redirected to the thank you page and the file you just uploaded will automatically be connected. For the sake of this demo, I'm going to click Get Visitors to Book a meeting. Next, I'll be asked to enter in the landing page description and the information that I add here will help generate the content and images for the landing page. So let me go ahead and enter in a description and then similar to what you saw Earlier. So within this flow you can select the if you have multiple domains for the portal, you can select the domain that you want to publish on. As Amanda demonstrated earlier, you can also select the ideal customer profile. This will help us cater the landing page to really reach the audience that you're looking to reach. And so for this use case, I'm going to select Data Analysts and then I can also select the content language. So depending on your portal, we will default to the language that your portal or your domain is set to. But you can also change language depending on the target audience as well. So all this information is being taken into account when we generate your landing page. We're also leveraging your brand, voice, company name and industry behind the scenes to help inform that generated landing page as well. So I'm going to go ahead from here and click on Generate Landing Page and we'll just wait for that to load. While this is loading, I do want to mention that as Amand mentioned earlier, this is currently in public beta for Content Hub Professional and Enterprise Portfolio portals. You can opt into this beta with access. You can opt into this beta for your portal and users with access to creating landing pages will be able to use this specific flow. The same permissions that Amanda demonstrated are applicable here as well. And so as you can see, for my use case of getting users to book a meeting, we have a meetings module added to the template. We auto pull in the logo that is set to the brand kit and all of the text on the page was generated based on all the information that we provided earlier. This can all be modified. So you can of course change out any of the text, can change out any of the images. The images and any of the CTAs on the the buttons on the page are also generated using AI and then from here you can go ahead and publish the page. I do also want to mention that within the settings we're also auto generating the page title, the URL slug and the meta description of the page. This can also be modified, but this is another thing that we're doing to help it much help make the content creation process much easier and faster as you're generating your landing page. All right, I think that is all that I have for you. As mentioned earlier, this is available in public beta for Content Hub Pro and Enterprise portals. And with that I'll check the chat and the Q and A for questions and then Lucas, I'll pass it over to you next.

That sounds great.

Thank you.

Can everyone hear me?

Yep.

Beautiful. Let me go ahead and Share my screen here.

Give me one sec.

All right, so can everyone see three tabs at the top says Podcast in the top left corner? I assume the answer is yes, but yes, let me know. Beautiful. Hey, everybody. My name is Lucas Pleva. I'm a group product manager for Content Hub, and I'm going to give you a quick demo of how we can use the content agent to create podcasts. So, as Amanda and Kelly have said, you can use the content agent to create a number of content types, from case studies to landing pages to blog posts. And I'm going to talk about the fourth content type, which is podcasts. So we know that podcasts are an increasingly popular way to share thought, leadership, build brand awareness, and just generally connect with your audiences. But it takes a lot of work to produce a podcast. You have to write the script, you have to record it, then you have to create a transcript, you have to create show notes, you have to create artwork, you have to generate an RSS feed so you can get it into Spotify and Apple Podcasts. It's a lot of work. And the content agent can help automate the majority of all of those tasks. So I'm going to show you how it works. So there's two ways to access the content agent and have it create a podcast for you. The first one is to go into the podcast tool. If you didn't know that we have.

That, if you have content we're getting, it's pretty small. Can you zoom in a little bit?

Yeah, definitely.

I suspect you have a pretty large monitor.

How is that any better?

Much better.

Okay, beautiful. Beautiful. So from here, let's talk a little bit about podcasts. So two different ways to get here. Number one, if you have Content Hub Professional or Content Hub Enterprise, you can click on Content and then just click on Podcast here. That will bring you to this screen. And then you can just click the Start with AI button in the top right corner. Alternatively, as with the other content agent jobs, if you go to Breeze AI, you go to Content Agent, and then you click on Launch next to create Podcast, you will be able to get access as well. Content Agent as a whole is in public beta right now, so once you opt into it from the product updates page, you will automatically be able to create any of these content types, including podcasts. So let's go read a podcast. All right, so the overall flow is pretty similar to what you saw for landing pages in case studies. We're basically going to ask you a few questions, and then we're going to use information that you gave us, as well as what's in your portal to create everything that you need to start podcasting. So in this case, I already have a podcast show created. A show is basically just a collection of individual episodes. The show is called Data Decoded Big Lytics Data Big Data Podcast. So I'm creating an episode here, and I'm going to call it the Impact of Artificial Intelligence, specifically Machine Learning on Data Mining. Right. That's what I want the podcast episode to be about. Going to answer a few questions about my audience. Most of my listeners are in the United States, and I can also specify, as Amanda and Kelly have shown, my ideal customer profile. If I have created a Persona in my HubSpot, where I specify interests, where they work, etcetera, I can set that here. Then when we're creating the podcast for you, we're going to take all the information that you gave us about the target Persona into account. I can also upload any relevant files. If I've maybe already written an ebook about data mining or maybe I've done some brainstorming and I have a Google Doc full of notes, I can upload that into this flow. And then when we're generating the podcast script for you, we're going to take the information from the uploaded file and factor that in. This is a really good way to ensure that the script we generate for you isn't just like generic summary, what you would find on the Internet. It actually takes your unique perspective into account by, you know, factoring in any of the documents that you've already uploaded, you know, as part of this flow. So I have the topic. I may have uploaded some documents. I'm going to specify. This is for data analysts, and I'm going to click next. As we've done for case studies, we're going to help you brainstorm some titles for this episode. And we're also going to do a little bit of SEO keyword research for you using data from Semrush. So for each keyword, we're going to look at monthly search volume, we're going to look at ranking difficulty, we're going to look at, you know, the keyword intent.

Is it commercial?

So are users looking to buy? Is it informational? Are users just doing general research? And based on that, we'll give you suggestions for what the title of the podcast episode can be. So in this case, I'm going to pick. I don't actually like any of these titles, so I'm going to see if I can get a few more That's a little bit better. So, Beginner's Guide to Data Mining and Machine Learning. So that sounds good. So I'm going to select that and then click Outline. So here we're going to look at the information that you uploaded, information about the Persona, the topic that you specify, as well as the title to generate an outline, which will then be used to generate the overall script. So within this outline, we give you about, I don't know, a dozen or so different sections. This is about 36 minutes of audio. And you can do different things in this outline to kind of really customize it and tailor it to be unique to you. So if you want to add additional talking points that will be in the outline, you can do that. If you want to remove any of the talking points, if you want to change them, if you want to change the section. So, for example, if you want to talk about machine learning before data mining, you can change the order. This is really your opportunity to add any additional context that you want to be reflected in the final script. Once I have this outline all good to go, I'm going to go ahead and click Generate Episode. This takes anywhere from minute to two minutes. We are generating almost 40 minutes of audio here, so it's not instant. But the good news is as soon as we're done, we're automatically going to send you an email as well as an in app notification. So if you can't sit around for two minutes and wait for this to finish, you want to go cup of coffee or something to eat, you can do that. You're not going to lose your progress, and we'll just let you know when this process is finished. Since we're short on time, I'm going to jump ahead a little bit and assume that this generation has finished. So we're going to give you a couple of things. Number one, we're going to give you a full script based on the outline of the previous step as well as all the information you gave us. So here that you see here, all these sections correspond to the sections on the outline step. And we've written out about 40 minutes of actual script that you can either download and then record yourself. Or if you want us to actually convert this into very high quality AI audio, you can do that as well. We have about six different voices that are currently available right now. They do sound very human, much better than what you would have on a 1999 Windows PC. So it is pretty realistic, sometimes almost freakishly so. We're also planning to add more voices in the future. But right now we have about half a dozen voices. In addition to giving you a script and letting you convert that script into audio, will also generate the episode description for you, which includes information about what's covered in the episode, will generate artwork for you, and we'll also generate a full transcript. So once you have all of this and you can kind of customize it and tweak it and make it your own, you can go ahead and publish it. Once you publish it, what we would do is give you a couple things. Number one, an RSS feed that you can use to share the podcast on pretty much any popular podcasting platform. So that could be Spotify, Apple Podcasts, et cetera. You can also embed the podcast directly on your HubSpot as well as non HubSpot pages. So if you want to add like a podcast player where you can listen to the latest episode or maybe even listen to all the episodes that you published, you're able to do that. So again, the content agent can help you create a podcast end to end, basically give you a script and then convert that script into audio, or if you just want the script and then do the recording yourself that's supported as well. And then from there you can just share it with your listeners and take it from there. That's about it. I'm gonna check the chat for any questions and yeah, happy podcasting.

Amazing.

What a group.

Look at the timing. We have a minute and 10 seconds left in the hour. I'm amazed. Thank you all so much for sharing so much information, live demoing, answering questions. This is great. Thanks all of you who attended for showing up, coming to this event. Love having you all here. For the sake of all of you here on the call, I can export out the Q and A if you want to take them back to your team as data points for you, things for you to consider. But yeah, I think we will wrap up here and thanks for joining. Another admin hug and hope to see you all back next week. Bye everyone.

Bye everybody.

Sat.