The following AI (Artificial Intelligence) features are useful for investment research (and thus also for the OFIP development):
IR1
AI Freature: AI agents replacing traditional investment research analysts and asset managers – LangChain |
Feature Description: – Agents are simply AI algorithms; the core idea of agents is to use a language model to determine a sequence of actions to take. – Agents may use various language models as a reasoning engine to determine which actions to take and in which order. – Agents may use superhuman amounts of information to perform their actions and may be able to reason better than humans by breaking down tasks into a series of smaller ones, and even include humans in the loop. – These agents may be capable of performing the tasks of investment research analysts, fund managers, and OFIP interns better than their human counterparts. |
Relevant Video(s) / Code(s) : Harrison Chase – Agents Masterclass from LangChain Founder (LLM Bootcamp) code 1 code 2 https://arxiv.org/abs/2304.03442 |
IR2
AI Freature: Agents significantly outperforming traditional investment research analysts and asset managers – CrewAI, AutoGen |
Feature Description: – Agents can perform many tasks almost instantly, like read long texts and summarize or understand the essentials and subsequently take action. These and other analytical tasks may be done by agents in seconds whereas humans would take hours or days to perform them. – Groups or armies of numerous agents, each having a different specialty, may be used cost efficiently. In general, only two types of developer may be needed to create these agents: Programmers to code the tools with the functions that the agents will need, and Prompt Engineers to retrieve from LLMs the relevant information and know-how. – Agents require very little supervision. In fact, GPT-5 is expected to be AI agent-focused, where agents would have the ability to perform tasks autonomously. – In the OFIP virtual assembly line’s complementary verification, each intern may represent an AI agent using the same tools but a different LLM. |
Relevant Video(s) / Code(s) : CrewAI Tutorial – Next Generation AI Agent Teams (Fully Local) / code AutoGen Tutorial 🚀 Create Custom AI Agents EASILY (Incredible) / code |
IR3
AI Freature: Agents being the ultimate competitive advantage of Wall Street firms – AutoDev, Devin |
Feature Description: – It has been proven that by using agents GPT3.5 may give better results than GPT4. – The future of AI is agentic: The competitive advantage of organization will likely be driven by who can build the best agents. Since more and more of our work will be performed by agents, whoever will build better agents will win. – LLMs will be able to open apps, create files, use the terminal, install new programs, fix bugs .. and much more. – Operating systems have evolved from Command Line Interface (CLI) to Graphical User Interface (GUI) and will now likely evolve to Autonomous Agent Interface (AAI), which is an operating system that gives complete control to an Agent – making AI many times more useful. |
Relevant Video(s) / Code(s) : Microsoft NEW AI Agents ARMY Is Here! Fully Autonomous SOFTWARE DEVELOPERS (AutoDev) / paper AutoDev: Opensource Devin – Personal AI Coding Engineer! / code |
IR4
AI Freature: Text Summarizations of various types and sizes of documents being one of the main tools used by agent – LangChain |
Feature Description: – Annual reports, new releases, due diligence documents, etc. may be automatically summarized with LLMs and thus make the job of analysts, bankers, traders (and any readers, in general) more efficient. – More importantly, these custom summaries may be used by agents to perform the same tasks as the above humans, such as trading decisions, and outperform humans. |
Relevant Video(s) / Code(s) : 5 Levels Of LLM Summarizing: Novice to Expert / code |
Some other Investment Research AI tools
Analyze SEC Filings using LlamaIndex question answering
If you would like to learn more about this work, please contact us at info@intelligenxe.com