Loading video...

Video Failed to Load

Go Home

Stock Analysis with AI Agents, using CrewAI 🚣 Using tools + LangSmith, running on Replit ⠕ A researcher, an analyst and an investment advisor Agents come together to built a recommendation on any stock out there ✨ lmk if you like and want more of this! 😉

128,808 views • 2 years ago •via X (Twitter)

57 Comments

søren's profile picture
søren2 years ago

@LangChainAI @Replit awesome. stock analyzer bots were some of the first things i built + hosted on replit in 2020/2021. @notjustinshaw throwback to the days of scraping SPACs on and gpt-2 summarization.

Alger's profile picture
Alger1 year ago

Looking to invest in the Enablers and Adopters of AI? Consider an actively managed fund investing in companies actively involved in developing and implementing AI technologies.

Mayo Oshin's profile picture
Mayo Oshin2 years ago

@LangChainAI @Replit Good job 👏 how can this approach be applied for general research on any topic?

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit Pretty easy actually, just a matter of updating the agents and tasks, so it would be only updating prompts, maybe adding / removing agents depending on what you want. I’ll share another example this week of a trip planner crew as well with similar setup

Richard's profile picture
Richard2 years ago

@hwchase17 @LangChainAI @Replit Code?

João Moura's profile picture
João Moura2 years ago

@hwchase17 @LangChainAI @Replit dropping it next week, probably Tuesday together with a couple articles

Juancho's profile picture
Juancho2 years ago

@LangChainAI @Replit I did something similar to this but the analysis is still too basic in my opinion, specifically if you really want to do a serious investment. I was thinking to probably create 1K agents to bring this to the next level

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit I think it’s maybe just a matter of having specific crews for what kind of analysis one would want, I tried to make this more generic, but if you want go deep and broad multiple crews are the way to go

PaulTheBully's profile picture
PaulTheBully2 years ago

@LangChainAI @Replit Is there a link to this tutorial?

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit Not yet, I will be posting in the next couple days together with a couple articles

Pete Sena's profile picture
Pete Sena2 years ago

@LangChainAI @Replit Super cool use of AI agents

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit right? Was pretty fun to build as well, but I have more cool examples to drop later this week 😎👉👉

Dave Hayes's profile picture
Dave Hayes2 years ago

@LangChainAI @Replit Certified dope.

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit 🫡 I’ll keep it coming

Justin McCarty's profile picture
Justin McCarty2 years ago

@LangChainAI @Replit What is crew ai

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit A thin framework to build groups of AI Agents ✨

João Moura's profile picture
João Moura2 years ago

idk if you saw this but another cool example similar to what we talked the other day

Evil Genius's profile picture
Evil Genius2 years ago

@LangChainAI @Replit Currently building something similar… Integrating into our platform…

Pablo Arango's profile picture
Pablo Arango2 years ago

@LangChainAI @Replit Hello, what is the software you use to record this kind of videos ?

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit Screen Studio, pretty swett!

Jan's profile picture
Jan2 years ago

@LangChainAI @Replit Thanks for video. Really good explanation what's going on.

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit Glad you linked Jan! So many great examples of things to build with AI Agents

Dog LLM's profile picture
Dog LLM2 years ago

@LangChainAI @Replit where’s source code

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit Dropping in the few days together with a couple articles 🫡

João Moura's profile picture
João Moura2 years ago

check this one out, I think you might like it 😉

Troublemaker's profile picture
Troublemaker2 years ago

@LangChainAI @Replit Any code for this Stock Analysis AI Crew? Sounds like fun.

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit will be sharing it later this week on the 21st together with a couple articles

Jason Ferrell's profile picture
Jason Ferrell2 years ago

@LangChainAI @Replit Cool, I built something similar using OpenAI’s assistant API. I found OpenAI to be rather slow, >1m per task run in GPT-4-turbo. Looks like you cut the demo for processing - how long do your agents take to complete their task runs?

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit The whole thing takes around 5-10m, but that's doesn't bother me as much as I face it as basically time back for myself to do other things as it runs. That said smaller local models would be faster but lower quality. You can use them through @OLLAMA super good for simpelr tasks

Muratcan Koylan's profile picture
Muratcan Koylan2 years ago

@LangChainAI @Replit Great Job! Working on a similar project and CrewAI is really helpful 💫

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit Thanks 🙏 I really appreciate hearing back from people using it ✨

Preet's profile picture
Preet2 years ago

@LangChainAI @Replit Good work brother.

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit Thanks buddy! Was really fun to build!

Dr Shrivastava A(Generative AI Consultant)'s profile picture
Dr Shrivastava A(Generative AI Consultant)2 years ago

@LangChainAI @Replit Super

Quang Dinh's profile picture
Quang Dinh2 years ago

@LangChainAI @Replit Great Job, I hope can try it.

João Moura's profile picture
João Moura2 years ago

@QuanginhNgc1 @LangChainAI @Replit Glad you liked it! CrewAI is opensource so feel free to give it a try, this example will be open sourced this week as well together with a couple articles I’ve been working on

Farid Safi's profile picture
Farid Safi2 years ago

@LangChainAI @Replit Good. Which software do you use for recording your camera ?

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit Screen studio, pretty cool

Jeff Burke ⠕'s profile picture
Jeff Burke ⠕2 years ago

@LangChainAI @Replit This is cool! Is the Repl public to fork?

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit Will publish this week together with an article 😉

Daniel Kempe's profile picture
Daniel Kempe2 years ago

@LangChainAI @Replit 👍

Dariel Noel 🏆's profile picture
Dariel Noel 🏆2 years ago

@LangChainAI @Replit Wow😍 how do you ensure the accuracy and reliability of recommendations from these ai agents?

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit Usually I'd measure this agains a testing set but a lot of this is built on live data, so I guess the best approach could be to have the report include all references for the data so you can double check if needed, if you include it in the prompt it should be able to do it

Fabian's profile picture
Fabian2 years ago

@LangChainAI @Replit this is awesome! great job with this chain.

The Content's profile picture
The Content2 years ago

@LangChainAI @Replit @memdotai mem it

推文点赞|网站:mf8.xyz's profile picture
推文点赞|网站:mf8.xyz2 years ago

@LangChainAI @Replit 🔘🔳🔲,northerners are all cowards. one or two battles will end the

stagflacja's profile picture
stagflacja2 years ago

@LangChainAI @Replit Very cool. What takes the longest in this process - getting openai chat responses / generating embeddings / getting additional data from the internet? BTW is this app related to your framework, or just coincidence :)

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit depends, because it's a non-linear process so sometimes it might do many embedding searches others not so much, but overall the OpenAI responses take the longest. Just a coincidence haha

hindu_interest's profile picture
hindu_interest2 years ago

@LangChainAI @Replit how to use this, website link please. Thank you

Danish Khan's profile picture
Danish Khan2 years ago

@LangChainAI @Replit What can be done to make the UI better?

Attila-IBS's profile picture
Attila-IBS2 years ago

@LangChainAI @Replit Has the stucked loop issue been fixed? I like this project, so I'd like to keep testing it, especially with Mixtral 8x7B

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit It seems that stuck loop was me putting together a very bad readme example I didn't really tested 🙃 I updated the test but still didn't close the issue as I wanna add guardrails to prevent that. I'm super curious to try it with Mixtral as well! Thinking about fine tuning it

DongNan's profile picture
DongNan2 years ago

@LangChainAI @Replit .

▓▒░cнroмυн░▒▓'s profile picture
▓▒░cнroмυн░▒▓2 years ago

@LangChainAI @Replit How much did it cost for generating this?

João Moura's profile picture
João Moura2 years ago

@LangChainAI @Replit seems it was < $1 but I'll run more tests and report back

José's profile picture
José2 years ago

@LangChainAI @Replit @NotionAddon

Ryan Lisse's profile picture
Ryan Lisse2 years ago

@LangChainAI @Replit @memdotai mem it

Related Videos

New Course: ACP: Agent Communication Protocol Learn to build agents that communicate and collaborate across different frameworks using ACP in this short course built with IBM Research's BeeAI, and taught by Sandi Besen, AI Research Engineer & Ecosystem Lead at IBM, and Nicholas Renotte, Head of AI Developer Advocacy at IBM. Building a multi-agent system with agents built or used by different teams and organizations can become challenging. You may need to write custom integrations each time a team updates their agent design or changes their choice of agentic orchestration framework. The Agent Communication Protocol (ACP) is an open protocol that addresses this challenge by standardizing how agents communicate, using a unified RESTful interface that works across frameworks. In this protocol, you host an agent inside an ACP server, which handles requests from an ACP client and passes them to the appropriate agent. Using a standardized client-server interface allows multiple teams to reuse agents across projects. It also makes it easier to switch between frameworks, replace an agent with a new version, or update a multi-agent system without refactoring the entire system. In this course, you’ll learn to connect agents through ACP. You’ll understand the lifecycle of an ACP Agent and how it compares to other protocols, such as MCP (Model Context Protocol) and A2A (Agent-to-Agent). You’ll build ACP-compliant agents and implement both sequential and hierarchical workflows of multiple agents collaborating using ACP. Through hands-on exercises, you’ll build: - A RAG agent with CrewAI and wrap it inside an ACP server. - An ACP Client to make calls to the ACP server you created. - A sequential workflow that chains an ACP server, created with Smolagents, to the RAG agent. - A hierarchical workflow using a router agent that transforms user queries into tasks, delegated to agents available through ACP servers. - An agent that uses MCP to access tools and ACP to communicate with other agents. You’ll finish up by importing your ACP agents into the BeeAI platform, an open-source registry for discovering and sharing agents. ACP enables collaboration between agents across teams and organizations. By the end of this course, you’ll be able to build ACP agents and workflows that communicate and collaborate regardless of framework. Please sign up here:

Andrew Ng

105,261 views • 1 year ago