Introduction

Mine's gonna be a lot shorter. So not to say that, mine's a lot less dense.

What I'm kind of trying to hope to do is what you see here.

Understanding the Audience

Now just before I start and I show you what's going on here, how many of you guys take notes or have a note taker on calls that aren't technical people but also are technical people? I guess everybody who,

And just out of the room, who here is in sales or marketing or? Okay, we got a lot of salespeople.

So just a quick question.

The Purpose of Transcripts

Just from the room, what kind of purpose do you have for transcripts generally? What are you using them for? Do you just use them for summaries or what kind of purpose does it serve your organization? Anybody?

To save time. To save time, just kind of summarize the notes, kind of... To make clear.

We feed them to reseller guides and projections handling summaries. Okay, so building kind of FAQs.

That's kind of what my talk is about.

Introduction to Blue Dot

I use a tool, so some of these, this is a little bit more, because every one of us is going to use, this is Blue Dot. So one of the cool things about it is it's got a lot of features, and I really like it for the one reason that it's able to

not only record the calls, but you can also do screen recordings. But that's some of the things, but what it's very useful for is it's got a very good translation in it.

So it's actually, you can automate a lot of these things and make, I'm not going to go into too much into it because what I'm showing you here is my general flow.

Leveraging AI in Sales Processes

I work in sales. I use dust.

One of the things that you might see here, for example, is we have a talk here with a potential client that we had for about a sales discussion. What I have is kind of a survey showed you based on some transcripts or later, but based on standard operating procedures, which is basically who's the client we're talking to.

Do we understand who we're selling to? Commonly known in jargon as the ICP.

So what you're seeing here very briefly is just a basic transcript that I had taken from a call that I'd taken.

All of this, of course, has been asked. Everybody knows that their information is being used.

And what I have set up, there is no workflows here. This is just, think of this as a new way of maybe working.

AI Tools for Transcripts and ICPs

Has anybody here used anything like Copilot or even in Gemini? Could be Microsoft, could be... Almost everything, I think.

Yeah, so you guys do realize sometimes when you look at, you might have, even when you go into Gemini, you click on the button, a new pop-up shows up and you have like a chat window within, let's say when you go to Google Drive, right? The reason why I'm not necessarily using Google Drive here is because, or Google, is because in this particular case, there's a feature in Dust that allows you to get the information from the transcript. and allows you to take a picture of the transcript itself.

So what I'm specifically talking about here, usually when I have a call, I have, I'll just kind of call it up here, this is a Chrome extension, but it's kind of simulating that type of environment. So I'm getting more and more comfortable working this, and to be honest with you, I don't work with anything else anymore.

But what the feature is essentially is I have built an agent, like I said, built on the ICP, what we're looking for in standard procedures. What I'm doing here is I'm getting the agent that I've built that I'm working with to read the page, but also from the transcript that has been created in Blue Dot. This doesn't always work, but you can take a screenshot of the full script and it's able to see that.

Based on the conversation, does this client meet our... How many people here have had a sales manager go, hey, did you qualify them properly? How's your discovery?

Right? Awful, right?

Probably you're just like, oh my God, I'm probably gonna get in trouble again.

Here, you've got an objective way.

Data Utilization in Organizations

One, you can share the transcript because everybody's looking at them, so I would suggest that as a best practice in your organization, really start thinking about how can we get more data ingested so we can look at that because when you get the data coming in, you may not necessarily know what to do with it right away, But as you start building out your AI into your organization, you'll start seeing that this type of information is extremely useful for all sorts of things. So that's more generally speaking.

And two, it creates much quicker conversations because you don't have to explain to somebody what you were talking about. So in this case, for example, I had a conversation. I had, let's say, director of sales, CEO, whoever it might be.

How did it go with that discussion with Enki? So what you see here, sales helper, just basically built on the ICP. It's going through, looking through the criteria.

It's compared against conversation. And then it looks at the ICP. So let's say we've got a full profile looking at this particular thing.

You can see that they've got, it's also because the agent is also equipped with a very simple thing like scraping the web. So it does the research for you. You don't have to build it.

This is a much less or a lighter version. Shows the company role, the industry, pain points that they might be facing, what their buying triggers are. some concerns where their price sensitivity might be to them, and then some recommendations.

Streamlining Sales with AI Assistants

So here as essentially you've built yourself, developers have been talking about having a co-pilot for themselves, having the whole time the ability to use something like Cursor, make it super simple. Here, you can do that as a simple, really easy to use way to, in sales at least, have a way to understand

because we all know whoever here is in sales room, if you've got a pipeline and you think something's going to close, if you don't have information, you're not in a very good position.

Just to kind of quickly close this out, give you the information very quickly. You can talk about other things you may have to understand, but essentially you've got a full, easy, very quick way to deploy a tool that you actually understand what's going on within your organization.

It's a lot shorter, I hope it's kind of clear but I wanted to just kind of ask if people have any questions about anything that they've seen here or any comments or what do they think.

I just have a question, what LLM does for sales help reviews, do you know maybe? I built this on, this one particularly. I would say just in general, I built this one on Claude 4. I think it was more just because I like the way Claude talks to me. It's a very nice LLM. It's very polite. I'm Canadian, so I get my feelings easily hurt, so try to avoid that by using LLMs that are nicer.

But of course, looking at what it would be built on, It really depends. Of course, part of it is how you prompt it and what the tools are. Some are advantages and different things, but I found that when I tested a few, Claude seemed to give me the best results, particularly for reading transcripts and giving me results that are predictable in that sense.

Which additional information did you give it in this case? So it says it goes to Google Drive to read some documents. Was it only ICP description or a lot of different documents?

Creating Effective AI Integrations

In this case, what I have is there are what you might call, and this is kind of good best practice to have without coding. Whenever you have anything set up or plugged in, you want to create what... SOPs, but what that essentially means is if you have business practices, it's a really good practice, especially if you're going to implement AI the way that I did here. You want to write it down in a very simple way.

One, because you want to understand what your business processes are. Later, you can use them like here. In this particular folder on the drive, this is all very much so associated not only with just the ICP, but in this case, it also has company mission, it has what kind of mails do we send out, it has a very large depth of information. So think of it as talking with your set of procedures.

And once you have it documented, you can just add into the file. The reason why I prefer dust in this case is because it's very simple to use and anybody can deploy it. But you can do this by creating basic custom smaller GPTs that might be able to do that for you, for yourself personally, by doing it on ChatGPT, for example. Yeah, you can basically, I've done the same on ChatGPT.

You can create a custom GPT, effectively, with the other, the Atex version is better. It's more user-friendly, because it's integrated directly into the source. It's probably more user-friendly than mine, but I've got something similar on and all the .

So basically, it's trained on the internal documentation about what kind of client you're looking for. But at the same time, did you also train it on the actual conversations that you had with the clients for the actual prospect That is something that I have built. What I've done actually is I've gotten the conversations.

I use this case because I want to show how simple it is to take a transcript and use something. But what I usually do is I have another folder they will have attached to this in this particular file where it looks through summarized conversations based on a certain structure. So certain standard operating procedure this is what it is, this is what you look for, these are things that you want to know, these are things you don't want to know, and output in a certain format.

So we'll have a look at this, and generally what it will do is read that conversation once it goes into the folder, and then it will just put it together for you and be able to produce for you. This is quite unstructured data.

Practical Aspects of Implementation

I want to show you how quickly you can structure data, but you can, I would also, I would recommend this to developers speak of cleaning your data. So if you can create summaries that you get from your note taker, clean them up to a certain structure that you would expect that's repeatable because One thing Dust does better than others because it's got built in a way to segregate the files to pull certain things out of it because if you are using LLMs in this sort of scenario, you want to make sure it's just easy to see what it's looking for and it will give you much cleaner, faster and better output.

So it's just creating another folder and attaching it to the agent. And what are you using for transcription? This is a tool, actually, Blue Dot.

I use it. Personally, for me, it's great because here, for example, it can also do things in Polish really well. So one of the cool things about it is I can have this conversation.

It's able to transcribe things. It can do it from Polish, turn it into English, for example, in a very quick way. That will allow you to... If you want to talk to it in Polish, you have a transcript in English, you can use the same sort of information to get that sort of information. So feel free after if you want to know more details about it, I can happily tell you more about how that works.

Confidentiality and Compliance

What about confidentiality? I mean, like, there's a sense of conversation and does the person on the other side know that it's recorded and it will be, you know, It could be in the files in our company for a year, two, or more. Is he aware? How do you think about it?

Generally what I always say, and that's why I say everybody's got transcription software, from my experience, unless I'm talking to a German client who wants to talk about GDPR, it seems to be a hot topic for them even though it's, I think, but for most people, if you are having those recordings, I always just make it explicit so it has been said, are you okay with being recorded? If they say no, then obviously I can't record them. But in practical use, You're going to get one guy in a thousand say, and he's probably maybe works for some sort of intelligence services, I don't know, maybe, but most people won't care.

But this is not only recording. It's also transcription, and then it's put into the AI. So I'm curious if the person would agree if it will know that this data will be transferred by some AI system. So, because maybe he don't want to, but his data are flying somewhere in the cloud, so. This is why I used Dust exactly, actually, I didn't want to say this, because Dust is SOC 2 and 2701 compliant, and they take on all the risk, so I'd rather pay somebody to have a look into that, and so is Blue Dot as well with this, so. Yes, you're right, but I want to pass it on to somebody else, so I don't want to be held liable.

LLM Pro Accounts and Data Management

In any case, if you have a pro account of any LLM provider, your data are not used for training purposes by default, so they are not shared with... You can very often turn this off, right? Plus there's a button to turn it off. On the pro account, that's default. On any professional account, like a pro team, a GPT professional...

You will find, by the way, you will find this issue with Germans. It's a long-standing issue. They don't like to even pay with credit cards. So, you know, they're suspicious. Yeah.

Technical Implementation Details

I also had another question, whether you've recorded the video, the conversation yourself with a specific software? Yeah, I use this, well I'm not, like as we showed Sergey, he'd probably spin up a program to do it. But I have to use paid SaaS products. But I use this basically, so I think the result is maybe similar, so I'm getting the same effect on the end of the data.

And when you've recorded the video, do you need to upload it then to Blue Dot? No, it's already there. It's already there. You can even do something like Make, which is quite popular.

You can just have, I don't want to get too technical, but you can do a web hook where you might be able to push the data into something that might read through it, summarize it for you, create it, and then have a folder that it puts all the conversations into. It's quite popular, at least for sales organizations that I work with. They'll put the files into there, and then you'll be able to access them for let's say, let's say if you want to go Stefano's route where you want a PDF and later check this conversation, this particular bot you have, just upload the PDF and you're good to go.

So Bluedotter is the best software that records the whole conversation, yes? Yep, it records the conversation but it's a transcripting tool I would say is the best way to describe it.

I see, thank you.

Conclusion

All right. Now I know for other guys, we got pizza supposedly coming here in a little bit. Yeah, we'll have one quick talk because I know Stefano, you guys want to hear this one because we're all worried about what the heck is going to happen with humans when we're all eliminated by our AI overlords. So here you go. Let me just unplug there. And thank you very much, guys.