Framer AI AgentsDesign and publish professional sites with AI
Bob's Research, filings read for you
Bob reads the filings, so you don't have to.
11 followers
Bob reads the filings, so you don't have to.
11 followers
Bob's Research is an AI analyst who actually does the reading. Public filings are long, daunting, and often built to bury the serious issues. Type a ticker and he works through the company's latest SEC filing — every footnote, every buried risk — and hands back a plain-English rating you can act on.
Curious how you handle the more qualitative stuff like MD&A tone shifts or unusually buried risk language that doesnt show up in standard risk factors. Does Bob pick that up or is he mostly summarizing what is already structured?
@aydngdxm Appreciate the question–It depends on the tier. The $5 / Free read pulls the standard sections nad rates from those, that one is closer to summarizing whats already structured, but it still has does a 100% analytic read and flags dubious stuff with, in my experience, very solid reasoning. The $20 forensic read is what you're describing. The extraction pass is instructed to skip generic risk factor boilerplate and pull the buried stuff word for word. Covenant language, accounting changes, impairments, etc. Stuff that lives in footnotes on purpose. So at the end of the day, the rating is built from verbatim passages in the filings, not from summaries of summaries. This was a design choice to make sure these sort of things get flagged explicitly and are not buried.
how does bob handle filings that reference other filings, like when a 10-K leans heavily on exhibits or prior quarterly reports for context?
@zmraahmetomizk So now the forensic reads pull in all accordant filings like 10-K, 10-Q, and 8-Ks (filtered by item code, so exhibit-only and Reg FD filings don't cloud things). For exhibits since the material terms of those are required to be described in the body and footnotes of the filing itself–which is exactly the text Bob's forensic pass extracts verbatim–they're not specifically read unless the agent deems it relevant.
It would be really helpful to have a side-by-side view that compares Bob's current rating with how he scored the same company in previous quarters so you can see if red flags are getting better or worse over time.
@poyrazsucu89403 Good idea, so right now the forensic read is 10-K, 10-Q, and 8-Ks, and the past ratings aren't kept around as I wanted a novel read, but a yoy rating is a good idea for historical performance. I will add this as a tab on the forensic read where you can compare previous periods.
@poyrazsucu89403 live and working.
Does Bob pull live data straight from EDGAR when a filing lands, or is there a delay before the latest 10-K or 10-Q shows up in his analysis?
@kaanokqv It pulls live whenever you do the analysis. Every request checks EDGAR first — for the forensic read Bob resolves the company's current filing set (latest 10-K, plus the newest 10-Q and any material 8-Ks since) before anything else happens. Ratings are cached, but the cache is keyed to the SEC accession numbers of that exact filing set. So the cache can never hand you an analysis of an old filing: a new 10-Q drops and the next person to ask gets a fresh read.
Tried Bob's on a messy 10-K I had been avoiding and it pulled out a footnote about contingent liabilities I would have totally missed. The plain-English rating felt honest, not hyped, which is rare for these tools.
@cengizrakalar Nice! Thanks for using it and for letting me know, that was the goal, plain language with a capital preservation focus hence the tone.
Pulled up Tesla's latest 10-K and Bob flagged the supply chain risk I'd been scrolling past for twenty minutes. Plain English summary was actually useful, not the usual AI soup.
@dnez5nl Wonderful! Great to hear and thank you for using it!