TLDR: ArthaLens: Free AI Summaries, Guidance, Q&A for Indian earnings call transcripts to save research time. DODLA’s guidance for eg. DODLA Earnings Calls | ArthaLens
Hey everyone! Like everyone, I use ChatGPT for earnings call analysis. We are currently building a free platform that leverages AI to extract important datasets from company filings like Annual Reports, Earnings Calls, Presentations, etc. As an initially offering, we built a super simple tool (totally FREE) that uses LLMs to help parse through earnings call transcripts and pull out guidance, Q&A, summaries, etc. Nothing fancy, but it’s been quite helpful:
- Summarizes the key points from the call
- Extracts the Q&A sections (which are usually the most insightful parts)
- Extract the Forward looking Guidance
- Lets you create custom analysis prompts for specific insights you’re looking for. For eg. You can give it a prompt like “Find all mentions of supply chain issues” or “Summarize the key risks discussed” and get tailored summaries for each transcript. Here is a quick demo:
Here are quick pointers:
- Full Coverage: It covers all the listed companies.
- No halluciations: Includes citations for each data point for verifications.
- Automatically Updated: New earnings calls are automatically added everyday at 12:00 midnight.
- We’ve only added transcripts from the last 4 quarters 01-01-2024 so far for now.
- The whole thing is pretty bare-bones and has limitations and bugs since its beta, but it gets the job done.
Thought we’d share it here since I’ve been a long time lurker on this forum and have learnt lots from here. This is just a small MVP that we want to get initial user traction and feedback from. We’ll be adding more features that include more datasets that extract key information from financial documents
Feel free to reach out for queries, features or collaborating! Kudos!
Website: https://arthalens.com/DODLA/concall
10 Likes
With the power of AI, we feel LLMs make quality datasets accessible to a much wider auidence without hefty charges to premium data platform like FactSet, Bloomberg, Refinitiv, etc..
We’ve justed added two new features/datasets that are available on FactSet Workstation that would help an equity researcher industry generalist as follows:
Business Strcuture - Gives a complete breakdown of a company’s division, departments, products, etc. Obtained from a company’s annual report. Example: PAKKA Business Hierarchy - Verticals, Segments, Offerings, Products | ArthaLens
Risk Factors - Contains key information that are risk as considered as risk by management. Obtained from Management Discussion & Analysis section of annual reports. Example: PAKKA Risk Factors Analysis - MD&A Annual Report Insights | ArthaLens
Please feel free to share what you think is missing or what additional features you might need for proper equity research powered by AI models! We’d be happy to hear any future feature requests and criticism.
2 Likes
Even the summary page is impressive too. Have you checked this for any data issues wrt the transcript aka hallucinations? Even this basic functionality if it is accurate could be a nice starting point for free tier followed by more functionality in paid tiers.
Another functionality that could be added is asking pertinent questions to the transcript, in case you are using a live API to an LLM at the backend.
1 Like
Hey! Thanks for the feedback. Hallucitations are prevented by precisely providing citations to page since we understand factual accuracy is the most vital aspect of financial research.
Regarding your second question, can you please clarify what you mean by pertinent questions to the transcript? Currently we have a “Custom Analysis” feature that lets you provide a custom prompt to analyse any transcript for it. This allows for more custom summaries and answers from transcripts. Is this what you meant? If so, please give it a try and let me know how we can improve! 
