How to Build an AI Slack Assistant in 5 Minutes (Chatbase)

Dr. Violet Howard shows how to build a custom AI Slack bot using Chatbase in five minutes, no coding required. She scrapes NVIDIA's Wikipedia page as a data source, tests queries about the company's IPO and GPU products, then connects the bot to Slack via Chatbase's integration.
Chapters
Today we're going to be using Chatbase and absolutely zero code to build an AI assistant within your Slack channel that you and your team can talk to in order to get questions answered fast. Let's not waste any time. Click the link in the description to go over to Chatbase and let's set up our custom GPT. Once you're in Chatbase, it's a super, super simple interface. All you got to do is click on New Chatbot. So now we need to set up our data sources, which is just the information that the AI assistant is going to be having access to. As you can see, you can upload files, you could just type in some text. You could put your website URL and it's gonna scrape your website and pull all the information back. You could do question and answer chains, or you could do Notion. As you can see, it's so easy to connect different data sources to an AI assistant. In this case, we're just gonna be using a website. So I've pulled up NVIDIA's Wikipedia page here and I'm just going to paste this URL in here and all we have to do is fetch links. So now in the background, Chatbase will be scraping the Wikipedia page for NVIDIA. It's gonna return all the information and as you can see, It just got a ton of information back about NVIDIA. And as long as we stay within 11 million characters, we should be good to go, which is a ton of characters. Now all you have to do is hit create chatbot and we can start testing this thing out. Now that our assistant has been configured and trained on some initial data, we can start to look around at the other options we have. So we have the playground where we can test out the responses and how information's coming through. We have activity to see past chat logs. We have analytics where we can look at chats, messages, how it's being trained, if the answers are good or not. We can also go back to our sources and we can delete stuff, or we can continue to add more information here. Then we have different ways to connect. We can obviously do it to a website, but in this video we'll be embedding it within our Slack channel. And then finally, we have our settings where we can update the prompting and the model of the assistant itself. Before we start talking to our AI assistant, I'm just gonna update this prompt real quick so that it knows it's an internal AI Slack chatbot and that it will be talking to employees from NVIDIA. So pretty much I just told it that its role is an AI chatbot designed to assist NVIDIA employees to provide accurate, friendly, efficient responses, drawing from its knowledge base, which we gave it that Wikipedia page on NVIDIA. And then we kept the other stuff about we wanted it to, you know, never mention that it has training data, focus solely on providing helpful and relevant NVIDIA-related information. We can also change the assistant's temperature, which is just kind of how reserved or creative the responses will be. So let's just go with 0.6. Hit save, and now let's start talking to our assistant. So the playground is where we can test out responses. So I'm going to come in here and say, when and why did we go public? And let's see what we get. NVIDIA went public on January 22nd, 1999. Going public allowed the company to raise capital by offering shares of its stock to the public. This capital could then be used to fund further research and development, blah, blah, blah. So as you can see, it's getting information from Wikipedia and it's getting it back to us super quick. And finally, we have the option to rate the response as good or bad in order to continue to train our chatbot, or we could have it retry if we didn't like the way that it came through. Real quick, I'm going to ask a more specific question. So I'm going to say, what are the benefits of the GTX 1080? And we'll see what they say. All right, as you can see, it just spit off 6 benefits here. We've got Pascal architecture, high performance, a better memory, SMP, VR ready, and energy efficiency. So that's going to be super useful if you're within your Slack and you need a quick answer. Now we need to actually put this into our Slack channel, so we're going to hit Connect. We're going to go down to Integrations. So Chatbase makes it super, super easy, and we just have to hit Connect. Now we just have to give Chatbase permission to access our workspace. Congratulations, your Chatbase has been successfully added to your workspace. We are thrilled to have you on board. So now let's open Slack and test it out. Okay, now we're in Slack. As you can see down here in our apps, we have Chatbase right there. And in the actual channel, I'm going to say @chatbase and then ask a question like, what is up? Now it says you mentioned Chatbase but they're not in the channel, so we have to quickly invite them. So there we go, Chatbase has been added. It introduced itself. Now let's try asking it a question about NVIDIA. So all we have to do is hit and you can see that Chatbase pops up. So we will hit Chatbase and then we can paste in our question. Have we acquired any companies in the past few years? Let's send that off. And now Chatbase will be working to answer that question. As you can see, He's typing, yes, NVIDIA has been active in acquisitions over the past few years. One notable acquisition was Mellanox Technologies, which was completed in April 2020. So it's got a little information there and we could also go back and ask for more information, but now it's in our Slack. I can ask it questions, anyone on my team can ask it questions, and it's really gonna help out the efficiency of the team. So that was a super quick video showcasing how to get a custom GPT into your Slack channel in no time. There's so much other stuff in Chatbase that we haven't explored yet, like other integrations, looking at the analytics, prompting and training your chatbot. So definitely let me know if you guys want to see more. Click the link in the description to go sign up for Chatbase, and I'll see you guys in the next one. Thanks.










