FREE Tribe AI Open Source RAG AI Agent GUI Builder!🤖 Self Hosted LangGraph + Ollama + Web Search
In this video
- 00:00Introduction to Tribe AI: Low Code Tool for AI Agent Teams
- 00:42Overview of Tribe AI Features and Use Cases
- 01:24Tribe AI Deployment Methods: Local and Remote Options
- 01:54Sequential vs Hierarchical Workflows Explained
- 02:20Agent Skills and Integrations with APIs in Tribe AI
- 03:15Step-by-Step Docker Installation for Tribe AI
- 03:55Configuring Environment Variables and API Keys
- 04:29Running Docker Containers for Tribe AI Deployment
- 05:03Accessing and Using Tribe AI Frontend
- 05:31Tribe AI User Interface: Light Mode vs Dark Mode
- 06:26Adding and Managing Skills for Agents in Tribe AI
- 07:16Creating Teams: Sequential and Hierarchical Structures
- 07:43Using API Keys and Configurations in Tribe AI
- 08:16Tribe AI API Documentation Overview
- 08:59Sponsored Segment: Stride Agents AI Appointment Setters
- 09:36Hierarchical Framework: Creating and Configuring Agent Roles
- 10:05Setting Up Marketing Leader and Worker Agents in Tribe AI
- 10:30Adding Skills and Knowledge Base for Worker Agents
- 11:07Configuring Open AI, Anthropic, and Llama Models
- 11:49Customizing Agents and Roles in Tribe AI
- 12:29Testing Knowledge Base Queries in Tribe AI
- 13:09Demo of Knowledge Base Queries and RAG System
- 13:47Querying Recent Open AI News with Agents
- 14:28Summary and Comparison with Other AI Agent Tools
- 14:59Closing Thoughts and Community Invitation
Links from the video
Transcript
3,113 wordsWhat's going on everybody Josh pokok here and we've covered many AI agent orchestration platforms on this channel even recently we've covered a crew AI gooey which allows you to basically create crei agents within a nice gooey interface in today's video we're going to be taking a look at tribe AI a low code tool to rapidly build and coordinate multiple AI agent teams I'm going to show you how you can self-host this open source tool and connect it with open AI anthropic
Or even ol Lama let's dive right into [Music] it all right guys so all links covered in this video like always will be in the description down below so what is tribe AI so have you heard of the saying two mins are better than one that's true for agents 2 tribe leverages on the Lang graph framework to help you customize and coordinate teams of Agents easily by splitting up tough tasks among agents that are good at different things each one can
Focus on what it does best this makes solving problems faster and better all right so I'm not going to read through this all but what are some potential use cases football analysis trip planning customer service and many more and some highlight features of this tool persistent conversations observability tool calling retrieval augmented generation human in the loop open-source models integrate tribe with external application through their API end points to interact with your team easy development you can effortlessly get this up and
Running using Docker and then multi-tenancy all right so there's mainly two ways to deploy tribe you can deploy it locally with Docker which is the simplest way which we're going to cover you can also deploy it on a remote server all right you maybe using something like coolify VPS whatever the case may be we're going to focus on the tribe locally today all right a few main concepts are sequential versus hierarchical workflows and we covered this yet uh a couple days
Ago in the um crew AI gooey one as well so in a sequential workflow your agents are arranged in an orderly sequence and execute tasks one after another and then in a hierarchical workflow uh your agents are organized into a team likee structure comprising of a team leader team members and even other sub team leaders the team leaders breaks down the task into smaller tasks and delegates them into its members and then skills so this is the ability to equip your
Agents to interact with the world all right such as using .g go Wikipedia Yahoo finance and you can create and use different skills so if your skill involves performing a HTTP request to fetch or update data using skill definitions is the simplest approach and here are the descriptions for these keys so URL is the URL endpoint for the API call we got methods headers functions prams properties all that good stuff the typical stuff you would use when making an API call
Retrieval augmented generation so everyone's favorite so it provides its own r system I'll show you that in a second so when we upload different knowledge into tribe it will actually um embed it chunk it Etc and then using open source models like I mentioned so you can install AMA and use olama models pretty much any open source model you can think of on there so let's go ahead and Dive Right In you're going to want to click on this link to
Take you to the docker install right here all right and this is pretty much step by step exactly what you're going to do so you're going to start by get cloning this repo right here and then CD into tribes so we're simply just going to copy this all right and as you can see here I already did this but I just simply get cloned and then cded into tribe and now you'll see that I actually ran code space dot this is
Going to open it in VSS code and now in the repo here it says you can just run this command to basically copy the example environment variable file to a EnV file which you could do or you could just simply do that within VSS code either one works so as you'll see here we got our example. EnV file right here here and you'll see that we have all of our environment variables here such as our username and password our first super
User admin our password right here so you can see it says change this for a lot of different areas which it would encourage you to change this especially if you're deploying this into production um or on a server or whatnot if you're just testing it out you really doesn't matter too much you are going to want to put in your open AI API key here and if you want to use anthropic Claud as well you can put that in here as
Well or you can see here these are the different embedding models that it's using so you can see the list of supported models right here in case you want to change some of this and then Lang Smith you can use this potentially for llm observability postgress information right here and then that's pretty much it now they do say in the docks to basically run this Command right here which will get you a secret token um in your terminal and then pitch
Paste the secret token into everywhere where it says change this and then build and run the Docker containers so you're simply just going to copy this Command right here now before you run this you're going to of course want to make sure you have Docker installed on your system and it is open and running and then you simply just copy and paste this command into your terminal and you'll see that it will start pulling all the different containers and setting up
Tribe right here all right so all the different containers once you run this command should be up and running on your system and then these are the different locations where you can actually access tribe main one we're going to be using of course is the front end okay so once you initially open this up it will ask you to log in so once you log in um first thing I would recommend is going to your user settings the main thing I
Would do is change it from light mode to dark mode because you know who uses light mode anymore and then there's admin over here so you could add different users so email full name password uh is it a super user is it active Etc all right we have uploads here so this is upload management this is where you can essentially upload documents and it's going to embed it and chunk it and all that good stuff so let's go ahead and test
It out so I'm going to click add document you're going to have to add a name a description and then upload the file right here all right so I'm just going to upload this a2p 10 DLC doc uh PDF right here that I did in a video a while ago so let's just change this to a2p you could change the chunk size if you want or the chunk overlap I'm just going to keep it at that for now and let's click
Save next is skill management so we have skills here this is pretty cool this is what we talked about before you can actually add specific SK skills so by default the different skills here are ask human so you can see here description is ask a human a question to gather uh additional inputs and then we have Yahoo finance so get information from Yahoo finance news we have Wikipedia so search Wikipedia and then duck. go search so search is the web using
Duck Dogo now you can add custom skills here all right so it has a pretty nice way to interact like a little gooey fun right here with the Json schema so like say if I wanted to add headers right here you could go ahead add like headers like authorization like so um you know Bearer like Bearer whatever the case may be right here if you need authorization um function stuff so you could add functions like enter skill name description enter skill
Description so pretty much you know any sort of API that you would interact with you can add this here here for your agents to actually interact with it and then the main area once you have that kind of set up that you'll be using is the team section right so you can create two main teams like we talked about workflows right here is either sequential or hierarchical let's say we wanted to create a researcher team right here you just go ahead
Click save you see I have a marketing team and a researcher Team all right so if we go to marketing right here you'll see that we have four main tabs right here we have build chat threads and configure so if we go to chat right here we have chat where we can actually chat with our team we have threads where we'll see all of our thread conversations that we had with our team and then configures so this is where we can
Actually create API keys so API keys are used for authentification when interacting with your teams through HTTP requests learn how to make these requests from the API docs so like I mentioned before there are API docs right here this is um open API we got everything that you could actually use to interact with tribe AI um you know if you're hosting this on a server such as login users utils teams members and all the get post uh puts deletes Etc threads
Uploads API keys are you tired of pouring thousands of dollars into appointment cters only to watch leads slip away imagine having a team of elite sales agents booking qualified appointments for you around the clock no more wasted time on training no more frustration with performance and no more draining your budget on inconsistent and expensive call centers introducing stride agents AI high powered appointment Setters that work 24/7 never get tired and book appointments while you sleep trained on thousands of successful conversations
Are AI agents El performed human teams at just on10th of the cost join the ranks of businesses that doubled their appointments and booking rates in just a matter of weeks don't get left behind in the AI Revolution visit stride agents.com now and transform your entire sales process with Cutting Edge AI technology it's time to accelerate your stride with AI agents all right so since we're using the hierarchy framework opposed from sequential you'll see that we're able to have different agents side
By side in a hierarchy if we're using the sequential framework for example when we go over here to researching you'll see that the way it works is you just simply drag agents below and it will create agents below so it's going to go through all these different agents every single flow now when you are using sequential it is a little bit different you could have a require approval for a human in the loop right here so each time it runs a
Specific agent it would ask you for your approval before running for the next one but let's go ahead and go back to our hierarchy and kind of break down how this works so at the top right here we have a marketing leader so this is the rues so you can obviously change this I would recommend making this a little bit more in-depth or Advanced you can always ask chat GPT to help you provide uh Specific Instructions depending on what you're trying
To do and then you just have your role right here you have your backstory right here so you're a marketing expert manager um since we're using the hierarchy you won't be able to add skills or knowledge base to the leader um in the sequential you can actually add skills and knowledge based to the first one and pretty much all the other ones but uh you will be able to add skills and knowledge base in the hierarchy to the um to the
Worker agents all right so here is the provider so we have open AI anthropic um n o llama so we can have like llama 3.1 any open source model you could use like Claude Sonic Etc we have open AI so I'm just going to use GPT 40 you could put proxy provider then you could change the temperature as well let's go ahead and exit that out now you'll see we have different uh agents here so like I showed before you simply
Just drag this uh from the node right here to create a new agent here you'll see we just have some dummy agents right here I created and these are worker agents so since it's a worker agent you're able to add skills and knowledge base um this one is just a Wikipedia researcher and duct researcher so answer any questions you are given that you could give it a backstory and then skills right here so you could select any skills you want so
We have duck duck search and Wikipedia and then knowledge base you could add specific knowledge base a model you could add a model right here and then proxy provider temperature Etc and then same thing over here we just got another worker this one has just the knowledge base of the robotics right here and then news researcher so um this is has access to Yahoo finance and that's pretty much it so obviously you configure these specifically for a more um you know
Custom use case before we test it out I'm going to just quickly add another worker right here and we'll just call this one a2p and I'll add the knowledge base of the a2p right here we'll change this to gp40 and then we'll go ahead and click on save so now we have four different potential agents right here um now you could make one of these agents too like you could make this um a leader agent and then you could have multiple
Leader agents or worker agents as well let's go ahead and ask what is a2p I'm going to actually ask how to get approved for a2p because it actually just told us what the definition of a2p is right here based on the llm uh knowledge base data so let's see we should be able to use the rag right here so yeah now we can see marketing leader is actually um going to the a2p knowledge or the a2p agent right here which is
Using the tool of the knowledge base so how to get approved for a2p messaging and then we can see the knowledge base right here so we can see what is a2p blah blah blah blah blah we're showing the information from our specific knowledge based document that we uploaded using rag right here so we can see the different pieces that it's pulling right here we can see the score right here the relevance um so what is a2p within a2p wizard all the
Different information right here based on relevancy and then we get our actual answer right here to get appr proof for a2p particularly for a2p 10 DLC in the US register for a2p uh select campaign user consent compliance all right so that's our answer there if we go ahead ask it about Astro robotics you can see here that it should query the knowledge bases again for astr robotics similar sort of thing all right now we can go ahead and ask it recent
Open AI news 2024 all right so we can see here at first it queried it tried querying the Yahoo finance news and it said there was no news found for the company um using the open aai ticker and then it actually used the uh duck ducko researcher agent right here and we got that open AI plans to launch Orion its next Frontier Model in December you know the Verge learn blah blah blah blah blah and we got the Wikipedia search right
Here or orine model launch training new AI model Chain of Thought reasoning right here so it will you know if one agent doesn't work lots of times it will ask the other agent all right guys so pretty cool stuff um I think you get the gist of it this is another AI agent gooey type um you know tool right here now opposed to the crew AI goey this one is more for a lang graph and you can pretty much get creative
Uh build custom teams use sequential use hierarchical add different tools add different knowledge using rag but let me know what your thoughts are in the comments down below about tribe AI have you tried it yet are you going to try it how do you think it compares to some of the other AI agent framework platforms that we've covered on this channel if you guys are new to the channel we upload videos about AI Marketing sales business growth so if you're new
To the channel you like that type of content make sure to comment down below like the video and subscribe to stay up to date with the uploads and then also too guys if you haven't already joined our free community stri community.com I'll leave a link down below to our free Facebook group and Discord Channel and then if you guys run a business and you need help with Systems Operations marketing paid ads traffic or implementing AI agents AI cool callers Ai call
Centers appointment centers into your business streamline your lead flow and sales process then book a call down below at executives.com apply to see if it's a good fit other than that guys I will see you in tomorrow's video keep hustling keep grinding and of course guys accelerate your stride take care