Empower Your Business with B2b Sales Prospecting Tools in IT Architecture Documentation

How to create outlook signature

hey everyone in this video I'm gonna share some of my insights about what I have learned so far messing around with the chat GPT specifically 3.5 uh and gpt4 in sales prospecting use cases and so I'm just going to share a couple insights of how we should start thinking about how to use artificial intelligence and sales prospecting and then I'm going to show some examples of how you could use artificial intelligence and so the biggest thing that I want to cover is in my opinion there are three major use cases to use artificial intelligence in sales prospecting the first one is to clean and categorize unstructured data so what I mean by this is if you have a data set that is pretty messy and you can all you need is to pull one's Insight out of that data set and then bring it into another column or bring it into another more usable format artificial intelligence is great at doing this when a regular expression won't work or when just splitting columns aren't going to work because they're not all by the same keyword or something it does a really good job of doing that the Second Use case is to summarize or infer decisions from inputs and so using somebody's LinkedIn summary recent news about a company um you know there are LinkedIn posts uh some of those kinds of things are um a phenomenal use case inside of clay using artificial intelligence because now we can create these custom Snippets which we're going to get into and the third major use case is creating a full email to reach out to prospects using artificial intelligence and my recommendation the reason I mention this is because I see a lot of people trying to do this and while this is a use case for GPT I actually do not recommend anybody who does this and the reason why is because when you do this artificial intelligence starts having too much variance in the output that it's going to deliver and there's just two you just lose control over the entire process and so when you think about well what am I actually trying to customize for what you're actually trying to customize for is that unique Insight that you want to bring to that Prospect at scale and that can usually be accomplished in Under 12 words you don't need to change your email completely in order to to get that across and so I don't recommend anybody creates full emails to reach out to prospects with inside of chatgpt I am a big fan of just using Snippets and bringing those insights into the Snippets um the next thing I would just want to cover is the difference between chat GPT 3.5 and gpt4 you know in my experience the difference between gpd3 uh The DaVinci model versus uh 3.5 turbo um 3.5 turbo or chat gbt is I would say I don't know 10 times better than the gpt3 with the DaVinci model and what also makes it amazing is the the fact that you can converse with chat GPT and in that way you can give it inputs and then you could show it hey here's what I want an output to look like and then you can give it another input and then it'll output uh data the way that you wanted it to and so that is a huge win for being able to stipulate the way that you want your data to actually be outputted now don't get me wrong gpt4 is very very exciting um and it's definitely better than chatgpt or if you want to call it 3.5 turbo um it is definitely better than you uh than those but for the extra cost and the uh you know running it and possibly having some down times because of the apis and everything honestly I don't see a huge difference when I'm using really good prompts in 3.5 with a lot of good examples and then when I'm using gpt4 as well gpt4 is more creative it does come up with a little bit better ideas and has better inferences but I honestly find that 3.5 is just as good um and so moving on from there I just wanted to cover how we should be thinking about conversing with chat gbt we're going to show this inside of clay but it's very much uh similar to the way if you're using the product you just use the the product and you just interface with it um and we're just going to take like a little break inside of clay from that document but so when you're conversing with Chachi BT what you want to be able to do is you want to be able to say okay I'm going to have four messages back and forth and so you're going to have a choice between the system the user and the assistant the system this is just the context that you're giving for the message and so you're saying oh I want you to be an assistant who's going to help me write a sales email you could also change this and say hey uh you are Dwayne The Rock Johnson trying to sell clay or your joint or you're like Conor McGregor trying to sell Salesforce right you could give it its context here of what character it should be playing I haven't seen it change the outputs too much but I just use the system to keep everything easy then you would come down and I would always start off with the user because when you jump into chat GPT you're the first one who starts the conversation so then we would start with user imagine when you're interfacing with chat GPT user is you whatever you're typing in that's what you're getting now the best part about the way that we have this set up is then you can have the assistant and you could stipulate okay when I give you this input I want you to give me this output and that is a really powerful upgrade for the way that we can interface with artificial intelligence and generative text in general so a lot of people ask uh should I be using gpt3 should I be using complete prompt all of these things honestly I think everything you could just switch over to converse with chat gbt chatgpt is amazing uh if you're using the API you've seen the big differences I move I've moved everything into Chachi PT now so let's go back to the dock and so here are some examples that we're going to be going over uh and I'm going to show the prompts in the way that I use it so the first example will be personalize an icebreaker about a company based on a company description we're going to clean some job titles we're going to infer along how long someone has experience in a role we're going to make some decisions about some data sets we're going to create ideas of how a prospect could use our product or service in their company uh Paul news for when a company was for when a person was working at an organization in their past history and asked if they were there or a part of the news infer what a person's focus is from their job title and for why a company would be looking to hire a role and how that affects their company mission and some other things and without further Ado let's jump into the prompts and see how we do it um okay so the first one is we're going to be creating a line the first line for an email basically saying I was on your website and it looks like you helped resolve every alert and stop reaches so basically this is for a company called critical start which is this is exactly what they do so the way that we set this up is we just had two messages and we said you're an A A you're an assistant helping me write the first line to a sales email and then just the user and then this is just very very easy using the input finish the sentence about what a company does this is the input and then we put the company description here complete my sentence in under eight words and use keywords from the input complete each prompt using my prefix here I was on your website it looks like you help pretty easy stuff this is a classic example of the way that I'm doing a lot of my prompts right now and so I'm going to break this down piece by piece and you'll see how we keep using this later on in the video and so if I have an input from an outside Source I always give it the context up front and I say using the input do something right I always say that and then I'll stipulate to it and I'll say this is the input and I put that in quotes so I'm making it very very obvious that this is the input after if I've established what the input is I give it the context of what I want it to do I want it to complete my sentence in under eight words and use keywords from the input basically I want it to be short I don't want it to go crazy and write you know a 10 paragraph essay off of this and I want it to use keywords from the input because if it only has eight words I want it to be very particular about the words that it's actually using and then here we're saying complete each prompts using my prefix here I was on your website it looks like you help this is how I keep control over the prompt so I know exactly what it's going to start with and then because I gave it this context I basically know what it's going to end with as well um and so let's see yeah so I always am basically stipulating an input stipulating the output of what I want it to look like and then stipulating what the prefix is so that I'm always in control of these things then for the Stop sequence I only used it on this one I put a period as a stop sequence and I do that here because I'm basically saying anytime you're about to generate a period stop and so that's how I definitely keep this line under one sentence because if it feels like it's about to generate a period then it just stops everything and it doesn't keep running over when I do my email sequencing I just put a period at the end and we're good to go and that's how I ensure that we're only writing one line about these things so that's how we write a first line on the company Mission another way that I use chat GPT is to normalize people's job titles if you were to reach out to this person and say hey vasu notice that you're the CEO founder of term Tegrity um they would know that that data is scrapes that that isn't a very good look same thing here hey Sam notice that you're the CEO co-founder of via Labs just it really doesn't look good um because you obviously took it from their LinkedIn profile so here we can see that we've normalized yeah let's just move this over we've normalized these titles so that we can use these titles when we reach out to people and so this is a good example of stipulating data for when I want that data so I had four messages so we're obviously going to stipulate something so I had I just gave it context at first then I said as a user clean these job titles and make them shorter and to the point and then I just gave it a bunch of examples that need to be cleaned and these I just came from the top of my head I could probably improve them more if I'd like and then for content for message three I fixed all of the titles the way that I wanted the titles to be fixed um so then I just stipulated to it exactly the way I want everything to look and then basically clean this job title and make them shorter and to the point did it again and now we have all these job titles normalized very very simple but this could really increase the chance of prospects replying if you want to use their job title in the output this one is a lot of fun and so what we're doing here is remember how I said we can use chatgpt to make inferences based off of data sets this is a really good one for that and so basically what I am trying to figure out is I want to know who has been in their company for two years and then they switched companies so basically I'm looking for somebody who uh they were at a company for two years and then they left and then they're at another company for two years and then they left and then they read another company for two years and then they left and so I'm giving it all of these people's start dates and then I it's outputting for me how many of those start dates so their past experiences how many of those were just a two-year Sprint before they moved on to the next one and so I said using the start and end today inputs go line by line and tell me if any of these dates are within two years of each other the output should be how many dates are between two years of each other the inputs are in you know this format if an input is empty just ignore it and keep analyzing and so this is just sample data right and so then all four of them are within two years of each other yeah that's true all four of them are within four years of each other and then this is just more sample data again just training it and teaching it the way that I want so this one this one this one and this one oh maybe it was four whoops I changed this just for the video but it was for whoops um so all four of these again were there and then we could just hit save changes and you can see from all these people's experiences we have uh you know this person was in a role three times in their past four experiences that lasted for two years and then they bounced to another company in the last four experiences of this person they have never just been at a company for two years and then bounced on to the next experience again this person did this person didn't really this person didn't really and and so on and so forth so you could give it a lot of data and just instead of manually looking at data you could have it make decisions which is absolutely phenomenal this is one of my favorite I think every company in the world should be looking at how they can use artificial intelligence to be able to do this um what we're doing here is we're essentially giving people ideas of how they would be able to use clay at their company and so what I mean by this is if you basically I'm taking a company description and then I'm giving a description about what clay does and then I'm saying Hey give them ideas about the way that they would be able to use the product if they were to start using the product right and so our prompts look like this and this is a very long prompt well mostly because we have the the description here but we're saying help me create uh creative ideas about how a company could use my product at their company the input I will give you will be the company description of the company create three ideas about how they could use my product my product is a sales automation tool that helps people build much better lists they can unify all of their data enrichments in one place and send better outbound messages which this can um obviously be improved the company description that I want to sell to is here and then we give them the company description boom boom this is gong right and then we finish up create three outbound campaign ideas for them and make sure to mention the types of job titles we would reach out to you and why we'd reach out to them be creative with the ideas but keep them under 12 words each so then I just gave it some ideas so that I could just show it hey this is what I want the ideas to look like and then we gave it a full company description and then we just let it rip and so we see here Target it manager seeking MDR Solutions improve alert uh resolution with automation reach out to csos of growing firms streamline detection and response processes and all of these things right and we can review some of these but they're basically saying like hey this is our product here's how you could use our product at your company because I looked you guys up one of my favorite ones to happen right now that I might even turn just that one into a video because that one's absolutely fantastic this one is a lot of fun what we did is we pulled their last uh we pulled from their past experience so their second or third experience I pulled a news story about their third experience company and basically did two things I summarized uh what the news story was about and I pulled in their company that they used to work for as well and so the way that we did this is we just basically said using the input summarize the news headline and complete my sentence with it this is the input keep the output under 10 words and use specific keywords from the input complete each output with my prefix notice you used to work at this company Were You There When blah blah so I noticed you used to work a critical start Were You There When MDR service launched for cortex uh oxium notice she used to work at surply Were You There When bootstrap Labs partnered with M particle Incorporated notice you used to work at the Business Council Were You There When joint Pavilion launched Rising Star class I think there's a really interesting one to be able to reach out to people and give some Nostalgia based on uh corporate events at their past companies as well and this one was also fun because we were doing two things in one this one is just very simple we were just uh inferring what people do based on their job title and what they find is important this could really be used as an input for something else later down the line so using the input helped me understand quickly what this person's responsibilities are within a company this is the input in under eight words explain to me what problems they may be facing in their role again down the line this could be uh interesting I would probably feed this into another chat CPT input to basically say okay this is these are the things that this person is concerned about in their company How Could my software help them uh in their company accomplish these goals so then we could do another inference thing um and then over here this one's a lot of fun too um what I did is I found these companies and what jobs that they're hiring for and then I basically made a AI generated line about hey I noticed you're hiring for this role and I and I would imagine you're hiring for this role because you want to improve X within the company so you can accomplish your mission the way we did this was using the two inputs to help me infer why a company would be hiring a person and put one as the company description input two is the open job role they are hiring for I just gave it to them write one sentence using keywords from input one about how hiring this person would help the company accomplish their mission keep the output under eight words complete my sentence using the prefix and all the stuff I saw in your job board that you're hiring for job title I'd imagine you're looking for someone to I'd imagine you're looking for someone to boost MDR Services adoption and awareness and that's for a director of growth marketing awesome awesome line uh virtual care manager I guess we would have to oh whoops we would just have to do a different run on that I saw your job board that you're hiring for an engineer in Chicago Illinois remote I'd imagine you're looking for someone to enhance media flies Tech platforms for sales teams awesome nope we won't read that one I saw on your job that you're hiring for a speech and natural language processing research engineer we should probably normalize these titles as well um I'd imagine you're looking for someone to enhance personalized adaptive learning experiences so when you're reaching out to these people about and saying oh hey like you're hiring for this job you could use artificial intelligence to say hey I bet you're hiring for this job because you're trying to accomplish something um and yeah so this is a lot of what I've been working on mainly I wanted to show this video so that you can learn how to structure your prompts a little bit better and how you think about artificial intelligence I think every campaign has an artificial intelligence input that you should be using and you just have to connect the dots on how well can you say okay I know what your problem is I did the research on your company and I can prove it and here's how we solve those problems is the the framework that I come to with all these things so yeah