thoughtly-ai

A no-code platform that enables businesses to rapidly deploy AI-driven voice agents for automating customer interactions and enhancing contact center operations.

Thoughtly AI

Hello hello. This is Maya with Lily the cleaner. Are you interested in $50 off your first house cleaning? Yeah sure. Great. May I ask you a few questions to make sure you qualify for this offer? That's fine. Are you a first-time customer? I am. How many bedrooms is your home? Three. Could you tell me how many bedrooms your home has? Three. Great. You qualify for the offer. Let me transfer you to a live agent who will book your first cleaning. Thanks.

All right. So what you guys just saw was an outbound sales agent created by Thoughtly that automatically places calls on your behalf. Today I'm going to walk you through how you can create one of those agents for your own business. You can check out all the simple steps so that you can create these on your own whenever you need to.

When we log into Thly, we first come to the dashboard where you can manage and monitor all your different Thly that are performing all the different use cases that you have. The first thing you're going to do when you create a Thly is navigate over to the agents tab. From there, you'll see all your existing agents and you'll navigate to create a Thly. You'll first be prompted to title the Thoughtly for this specific use case. We said we're doing an outbound sales agent, so let's call it an outbound sales agent. 

For conversation type, we are using this agent to increase sales for our business, so we'll create click sales. Then we have three options. We can start from scratch, basically build the conversation that we want the agent to have ourselves using our easy no code drag and drop. We can also import scripts, so if we have scripts already, we can put them in here. Or we can go to our Marketplace, that's actually not yet available, so stay tuned for that, where you'll be able to use different templates from other users from Thoughtly. 

So for today, we'll start out on a blank canvas, and this right here is our agent editor.

This is where we will create our agent. So what you see right here are our notes. The first node is a start node. This is where you will go ahead and add in how you want your agent to greet your customer when the customer answers the phone. So automatically we have in here thank you for calling thly. This is Tessa. Who am I speaking with? This is an outbound sales, so the customer is going to be answering the phone usually saying hello or this is Mary, who is this? Hi, who's this? 

So let's instruct our agent on how to engage the caller. To begin with, we'll say that we want to say, you know, hi, this is Tessa calling from thly. I saw you were interested in our platform. Can I ask you a few questions? A question like this is a great question to start with because it prompts the caller to give a yes or no answer, which is always the easiest type of an outcome to give an AI agent to deal with. So can I ask you a few questions? 

From there, we anticipate some outcomes. We anticipate that the caller might say yes, but they also might say no. They also might say not right now or who is this? Because they might not hear that she said she was Tessa. A lot of different outcomes are possible. These are possible things that the caller might respond with, so we list out those potential outcomes here. 

The next thing we'll want to do is basically for each outcome, we want to script out what the AI agent will respond with. So the way they respond is usually using a speak node. So again here we have automatically, it's great to meet you. Thanks for using thly. Let me know if you have any questions. But of course, we're going to need responses per outcome. 

So if the person was to say yes, I am available to answer questions, we want to add a speak node here and we will drag the yes to that node. This is how we want the AI agent to respond if the caller says yes, you can ask me questions. So if the caller says yes you can ask me questions. We'll have the AI agent say great and we'll say you know have you tried the platform out. And then the out comes again yes they have tried it no they haven't tried it they're not sure they don't really remember what we're calling about. Let's draft all these different outcomes that the possibilities out. 

And then again what we did before is they say yes we've tried it great. Let's add another speak node and from there what do we want them respond great I'm glad you have tried the platform out. What did you think you know or maybe not what did you think but you know do you have any questions about how to use it. 

And then again outcomes yes they have questions no they don't have questions. We can list out all the different possibilities so that's the yes tree that's a yes yes all the way down tree. And then let's just say you know they say you know no I don't have any questions. Well that's when we going to attach the no down to the end node and so they'll say okay thanks. 

So that's one of the basically trees that the agent will go through. So just like we just scripted the agent down the yes uh decision or the yes outcome. Let's now script it as if the AI agent called me said hi. Like we said this is Tessa calling from thle. I saw you were interested in our platform can I ask you a few questions. I'm busy I say no. Let's script that out that right there would go all the way to the end node because we drafted out the end node and said okay thanks. 

All right so what about if someone was to say not right now we would go ahead and attach that to an end node and we'd say similar to what we said before but I want to I want to change the the ending a little bit instead of say you know no worries we know you are busy we will try you back another time.

All right last but not least uh we have a outcome here where it's who is this. So Tessa calls she says hi this is Tessa calling from thly I saw you were interested in our platform can ask me a few questions the person didn't hear it is say who is this. So let's put in a rule here which is under rules you scroll down past outcomes and basically you say if the caller asks who asks you know if the caller doesn't hear the initial greeting and is unsure of who is calling repeat the initial greeting.

And so now we've just added a rule rules are very easy it's kind of for one-off situations when you want the AI agent to handle conversations in a specific way and we'll also be able to in a bit go up here and train Tessa on the top right so we'll see this rule happening in real time and we can make adjustments maybe she didn't understand what that rule meant she's not like what do you mean if the caller doesn't doesn't understand what I'm saying.

Training lets us go into the conversation test out the conversation see what she might respond in a specific situation and then we can continue to train her and coach her um on the ways we'd specifically like her to handle a call. So um now that she knows to repeat herself she'll then take them through the speak node the last thing we need to do is script out the remaining outcome.

So we have two more outcomes when she asks have you tried the platform out this person says no so we want to add a speak bubble for someone who says no I haven't tried it out and we'll say um you know we recommend you try out the platform to get a sense of you know how it might. 

Benefit your business. You do. You want me to send you a link to try it out and then again yes or no. So let's take the no to the speed bubble. 

And then this last one was I'm not sure. Have you tried the platform out? I'm not sure honestly. We're going to put set them down that same path too. We recommend you try it out. 

And then finally we have, you know, two more outcomes now on this. We recommend you try it out. Can I send you a link to try it out? Someone says yes. So then we will actually put in here an action node. 

And this action node is when we will tell Tessa to send a text message that has a link to child platform. And then with the no, um, they don't want to text. We can basically say okay, you know, go here to the end node. 

We can also take it here to this end node. We can also create a new end node that says all right, well no worries if you don't want to try it out. The other thing can do is say no and then we can direct them to the next set of questions. 

Do you have any questions then on how it works? That's probably the best thing to do. So let's send them back up here. And then I accidentally just deleted this action node so we'll do this here. 

All right, and then finally we have one more outcome to take out which is the have you, you know, do you have any questions on how to use it? Yes, they do have questions. 

So here in the rules, you know, answer any questions they have based on, you know, what the knowledge you have on thly. So what knowledge does the AI agent have on your business? That all comes into play through our genius feature which is right here. 

The genius feature is basically your ability to attach your database, attach your website, attach call recordings, anything that you're going to use to train your thought lead to understand your business. Like I'm sure you have policies, return policies. I'm sure you have other different things that you know maybe it's a list of. 

Products of catalog whatever it might be that you want your thly agent to be trained on. This is where you're going to upload it and this is how they're going to know all they need to know about your business. So that rule right there is going to be pulling from to tell them to say hey answer any questions knowing what you know about thly. That what they know about Thoughtly is right here in this genius.

All right, and then once they've asked question they've answered it, we're going to route that to the end node. So this is how you would create an outbound sales agent with thly. Now that you've built the AI agent, the next thing you're want to do is you're going want to train it. So you can have it call you to see you know what sort of experience it's you know it's going to have. How is it going to take these rules you put in? Is it you know understanding the different nodes? 

So you could call you, you could talk to it, you could try to break it. The other thing you can do just train them through this sort of kind of SMS visual I like to call it. But you can basically give it different outcomes. So we said hi this is Tessa calling from thly. I saw you were interested in our platform can I ask you a few questions? Yes great have you tried the platform out? Um no okay thanks. If you end up wanting to learn more about the platform please visit Thoughtly.

So we can see here that that's not going to read out correctly the website. So that's something that we want to know. Okay, we've seen that the website's not correct. We'll go back and we'll edit it in the speak node. 

One thing we haven't done yet, which you can do in the beginning or at the end of creating your whole AI conversation, is customizing the agent itself. So we titled the agent outbound sales agent. We chose sales as the use case, but we are given a defa a default voice as test. But we actually have a whole library of voices that we can choose from. 

So you can navigate through here. You can listen to the different voices. Um, there's English, Spanish, French, Indian. Uh, there's an Australian accent, British accent, Portuguese. Um, and you can also have the opportunity to use custom voices through 11 Labs. Uh, you will just have to talk to our sales team to get access to that. 

So this is Tessa's voice. Hey, my name is Tess. I'm very much looking forward to our chat. But we can also use someone like James. Hello, I'm James. Let's get started. It really depends on, you know, your brand and how you want your brand to be reflected. 

So we'll go ahead and we'll go with English. Um, and then the other thing that you can do is hop down here into presence. Presence is where you can choose the background audio. You'll notice that when we did that test call, we heard the background of typing. That call center noise is you can do that, or you can also do a city street, a coffee shop, the subway, or you can do, you know, no background at all, car interior, or more. 

So for today, we'll just go to none. And then you can also go ahead and customize the personality of the agent. You'll see here assertiveness level and humor level. So if you want your agent to be a little bit more pushy, you know, when we ask those questions of have you tried the platform, do you have any questions, we can increase that assertiveness level if we'd like to. 

Have softer we have balance balanced assertive yet kind for our medium or we have high which is the max assertiveness really you know catered towards those direct outcomes. We also have a humor level so that's serious or we can do so off. We can do serious and minimal for low. We can do balanced humor and professional for medium and then we can do you know very high humor so that they can they're a lot more py and it really again just depends how you want your brand reflected.

All right so let's go ahead and try testing out the agent that we just built. I've navigated open over to the responses Tab and I'll go ahead and have the agent call me. 

Hello hi this is Tessa calling from thly. I saw you were interested in our platform can I ask you a few questions? 

Sure great have you tried the platform out? 

Yeah great I'm glad you have tried the platform out. Do you have any questions about how to use it? 

Yeah I actually don't know how to add a team member to the platform so we can both build out the thly agent together. 

To add a team member you can go to the settings section and look for the team management option. From there you can invite your team member by entering their email. Once they accept the invitation they will be able to collaborate with you on the Sol agent. Does that help? 

Yeah that does thanks so much. 

All right so what you just heard was us testing out the agent for the first time through that responses tab. We can see the call summary. We can see um the different tags. We can also see the transcript here as well as the recording. 

Now that we've tested out the agent through the training Tessa option we've also done the call me option we can go into any of the noes and make any of the necessary changes that we need. 

Note, that there may be slight differences in text and video. Watch video for the original content.

Thoughtly AI Video