Show HN: Search and analyze millions of SEC filings with AI. | Hacker News Reader

Show HN: Search and analyze millions of SEC filings with AI.

publicview.ai

41 pointsashr_17 comments

Bad UX... "chat with thousands of filings" but right after: pick a ticker to start. Sure, click the Chat tab instead. Type a query. Roadblock: you have to log in first. After login, boom query wiped. I'm sure people won't use it for the UX, but UX matters...

Chat with "millions" of filings. As long as you select them one at a time.

This approach allows for better accuracy per response since the context is shorter and more precise. "Search and analyze" was the messaging but I get your point. Thanks !

Getting all the filings is a lot of work, too - you have to build an old load parser

Good feedback, will work on saving the user's query before the signing in.

FWIW, OpenAI uses SEC Filings as the basis for their Assistants tutorial:

https://platform.openai.com/docs/assistants/tools/file-searc...

I've seen hundreds of these types of apps, to the point that I used Claude to build my own similar version of this app with similar features 100% for free.

What is the moat / advantage here when I or anyone else can build my own and at my own will serve it as a SaaS if I want to?

I'm assuming data isn't the moat here either as we all get the data from SEC Edgar.

From my perspective, the moat in such cases would be how many users impressed (based on accuracy, specific pain point) and how fast were they impressed.

Part of "how fast" equation is the UX - build with zero friction - every second they gonna churn unless impressed.

Yeah, people care a lot about convenience and ease of use.

The advantage here is the ease of searching for a filing. It's more convenient than having to search the EDGAR database then upload the doc to Claude every time.

I have built something similar for my personal use.

The problem is that some filings especially the 10-K onces, can be very long (too long for chatGPT or Claude). Only Gemini can handle them properly. It can be quite expensive if a user chats a lot with a long filing.

Yeah this is true. But the context windows for most of the frontier models have been increasing over time so this will most likely be solved soon. It's good enough to still get a decent amount of value.

Who's the target audience here? Trading firms and funds have a comparable, likely far better system, already.

Many hobbyists and "algo traders" only care for actionable results. Going through millions of records manually, even with AI, isn't really helpful.

Maybe I am in contact with too many quants, but this isn't even 5% of what's relevant for any given financial decision. Find out what the other 95% are about, integrate them into the product - now you have Bloomberg Terminal, great.

Majority of the users are hobbyist value traders and traders at medium-sized firms that spend a lot of time going through 10k's and 10q's manually.

That's kind of what I was thinking... do you have an insight as to why they're doing that? For mid-sized firms there's a lot of modeling around risks: possible demand issues, possible interest rate rises, comparison to competitors - I am skeptical that this can cover all of it.

For hobbyists, I don't think that most of them strictly want to go through all these fillings. You could likely deliver "dumber" results, providing just bullet points on what's particularly interesting about a specific ticker.

Just as a heads-up, I am not attacking your product, it's just that the financial industry is very broad and you have to be mindful about your target audience. I have simply seen quite a few products that are "LLM with dataset X" fail because they don't integrate nicely with preexisting workflows.

Most people use this tool as an alternative to the official SEC website since it also displays the original SEC document alongside the chat interface. So you can think of it as a complimentary tool to other financial tools (like Bloomberg terminal etc) as opposed to a replacement. The pricing also reflects this.

To be fair, I didn't look at the pricing before leaving my original comment - $9 a month is indeed very well priced.

Reply on news.ycombinator.com