Depends on what people want. My contention is that staffers can use this to reduce their reliance on lawyers and that lawyers can use it to help their union clients. It will definitely help me when I am asked about specific contract language.
I worked directly at the SEIU international, the UFCW international, and an IAM regional office. Nowadays I contract with unions as an outside lawyer. Not saying this is some kind of groundbreaking improvement (unlike my NLRB Research bot, which I think is), but it's helpful and a nice addition.
Sometimes it's useful, sometimes it's not. It also may not be that useful when the charge is $125/hr ;). But how does having a contracts database trade off with these other things?
What's funny is the bulk of the pdfs currently in there come from a DOL database (the remainder come from trawling the internet). The difference between the DOL database and mine is that mine is full-text searchable. Please write a post about how the DOL is doing a disservice.
So read you know 10 of them and pick the one you think is good.
"filling vacancies" is a specific phrase that only occurs in 10% of the contracts, you can sure for just "vacancies" if you want to get more, you can use OR operators, this is the most basic search engine possible, using technically that existed decades ago
What robot are you talking about. It's a word search. Put in a word or phrase and it will show you ever CBA that contains that phrase. You then click the CBA and read it.
I know what they want because I get emails from them all the time and I actually do this work. That database provides something that did not previously exist, which is the ability to instantly find contract language to model your proposals on.
The LLMs fetch PDFs from the internet and guess at the employer, union, and expiration date name. None of that particular matters. What matters is the part where the pdfs are converted to plain text and them that plain text is made searchable. That does not involve an LLM. Just regular coding.
The research assistants only hit the MSPB Law and NLRB Law databases. If I can come up with a way to use an AI agent to make it easier to get useful info from the Union Contracts DB, I will. But I haven't so far. You just put words into the search bar and it shows you all the CBAs with those words
That's not true at all. Over 90% of them come from the DOL. Then every day, I have an LLM go out and find 5 or 10 new ones that are posted disparately on the web. What makes my database unique is that it can actually be used to find example language for real bargaining.
scans of contracts from the 20s that can't be converted to clean, searchable plain text is not useful to a union trying to find example language for negotiating a CBA today, you really just don't understand what problem this solves
The point of this database is to allow people to find example language. The employer and union fields are best-guesses based on the text of the CBA, but don't actually matter. What matters is that if you search for something like "Filling of Vacancies," those contracts pop up in the results.
Finding example language is a common issue labor lawyers and unions have, so much so that there are email lists where people ask whether anyone has example language for this or that all the time. As you can imagine, this is very useful for those people.
The point of this database is to allow people to find example language. The employer and union fields are best-guesses based on the text of the CBA, but don't actually matter. What matters is that if you search for something like "Filling of Vacancies," those contracts pop up in the results.
It's worse than that. They are taking my money!
I've been following this project for awhile with some interest. I think the real heavy lift here is creating the subject matter database, and making it usable enough that the llm can effectively follow citations and generate hyperlinks. It's very cool!
I've approved you
let's find someone who knows about the NLRA and have them prepare a set of questions and the answers they expect, and then I will run the questions through the bot and see if they match (this is how I have evaluated it of course since I know a lot about the NLRA)
let's find someone who knows about the NLRA and have them prepare a set of questions and the answers they expect, and then I will run the questions through the bot and see if they match (this is how I have evaluated it of course since I know a lot about the NLRA)
You should give it a try and tell me what you think
When this happens, what it actually does is say that some precedent says X while other precedent says Y. It is also able to pick up overturning of precedent over time and reports that accurately as well. I've seen it!
Yes it's a research tool. Indeed, it will only produce legal memos, not any other kind of document, like a brief. And it's absolutely the best way to start any particular research you want to get into. You can use the memo as a jumping off point if you want to do more.
the subagent writes a python script to find the exact quotation string in the .jsonl file, i watched it do it! what is the other point about propositions? You are saying i need to run it after the memo not before?
Since all the sources are already downloaded in advance and stores locally in .jsonl files, it is very straightforward to have it go through them
Yeah I've already got it working it looks like. Before it writes the memo, it writes this JSON file that lists every quotation and proposition and corresponding citation. Then it spawns a subagent with clean context window to verify each one (quotes deterministically, propositions in its LLM way).
Quotes can be verified deterministically because when it queries the database, it stores the JSON search results (which includes full text documents) locally. So that part will be easy. Proposition checks more fuzzy but will run it through them anyways.
Ok I am implementing this now. The thing already does a "citation audit" before it writes the memo. But that is to check that the links work. I will have it create a .json file with citation->proposition pairs and then spawn subagent to check those after the citation audit.
Yes I did something like this for my book to sort through the text after it was written.
To be clear, what my tool does is produce a research memo. Totally different situation if what you are doing is drafting something for submission to the NLRB or a court. This would never be submitted anywhere. If you were to write something based on it, you'd click through to the sources.
