Your Agreements Are a Database You Can't Query — Hiral Shah, Docusign & Sean Sodha, NVIDIA

AI Engineer · 18 min · 168 sentences · from YouTube's caption track

Each timecode opens YouTube at the start of that sentence. Line anchors (#s42) are the cue ids in the WebVTT, and every line carries its start and end seconds. All transcripts has every talk, and the whole corpus as one file.

00:01[music]
00:13Uh, welcome everyone to our session.
00:16Um, I'm Haral.
00:17I'm a senior director of products um at DocYsine and I'm joined by Sean.
00:21Hello everyone.
00:22I'm a product manager at NVIDIA.
00:23So today Sean and I are going to talk about a massive problem that every enterprise faces which is agreement data largecale agreement data.
00:34Agreements are a big part of any relationship any B-2B organization kind of goes through day in day out and a lot of that data is captured inside that agreements and it's very critical whether it's pricing tables a lot of it and it's all in a lot of different unstructured
00:53format and that's kind of what we're going to show is how docysine partner with Nvidia are fixing that on making that data available readable usable for a lot of our kind of organizations.
01:07So here is kind of just a quick map of our talk today.
01:10We'll start with just the stakes.
01:11Why does this matter?
01:12Why the scale is so large?
01:15And then we'll dive deep into the technical architecture of how we are approaching it, how we have tackled this thing, especially this document processing at at scale.
01:25And then finally, we'll cover what we have learned from our evaluation of all of the different models we've tried for different purposes.
01:33and share our learnings with you.
01:35So why like you know just to understand why this is a big problem when you think about docysine raise your hands how many of you have used docuign anyone who's employed probably the HR docs right so it's a massive
01:49scale everyone uses docy for us it's a massive engineering problem as well because just look at the scale we have 1.9 million customers who are paying us and a billion users
02:03what does that imply we process us a million agreements a day that needs to now structurize make it readable make it querable usable and you know in the past we've worked with Deote on a study and it says that there's $2 trillion
02:19captured in this agreement negotiated value that no one capitalizes no one goes back and gets that um data back right and why it's because they have to do a lot of human reading human reviews there's disconnected
02:32systems lot of manual workflow flows that are there.
02:36So that's kind of why docysine built AM an intelligent agreement management platform that takes the entire like applying in an AI first way the entire agreement life cycle whether you are creating agreements
02:49genai helps a lot with that whether you're negotiating to understanding and redlinining all the way to after signing storing and making a lot of insights from this data.
03:00So when you think about the challenges involved right in an agreement there is the unstructured data it could be a PDF it could be a PNG and what are people wanting to do is simple questions they can't get that that is the data that is trapped inside
03:17one agreement but also the whole corpus of millions of agreement 10 years 20 years of business has kind of put into that.
03:25So for agreements aren't flat they're also hierarchical in nature they are like you know one agreement governs the other the other kind of does something else so you always are needing
03:37lot of things to answer this question simple thing I'm sure you all are using a lot of right claude and all especially at your company's organization simple thing what did we contract for the total tokens right with claude no one knows that's captured
03:50inside this agreement in different forms and fashion so you need to be extracting this data to find things but also insights and push it downstream where you're tracking doing more things
04:00and when you analyze an enterprise contract a big set of things are captured what I call vital terms like pricing tiers the skues the information SLAs's rate cards they're all in table format
04:14now traditional document extraction tools or a generic VM BLM completely fail here like you know we've tried we've definitely done this because they're reading text line by line which breaks a lot of that concept within the table.
04:28A merge sells some things are not boundaried.
04:31So this makes a massive operational overhead for the downstream legal teams, procurement team, sales teams to get that queries and get that answers done and they spend hours and hours digging through this
04:43um to even just locate a basic thing.
04:45And that's kind of where we partnered with Nvidia and leveraged a purpose-built model like you know tool which is literally like you know making our architecture for table extraction
04:57take us to make and solve these complex use cases.
05:00We're trying with this we're making things scalable.
05:04We can understand it with the layout but also deliver really accurate results.
05:10And to share more about how we're leveraging the Neotron, I'm going to hand it to Sean.
05:17All right.
05:17Hello everyone.
05:18Uh so real quick on uh the Neotron retriever initiative.
05:22So for those who here knows about Neotron, you raise your hand real quick.
05:26Awesome.
05:27Uh so Neimatron is all about building worldclass open-source models and publishing the data sets, the techniques, uh the quantization approaches, distillation approaches, pruning approaches, every technique possible, blueprints to go with that, you name it.
05:41Um throughout the Neimatron portfolio, we have specifically Neatron Retriever, which is building embedding models, reranking models, and document extraction models.
05:50Um so real quick here we sort of our first initiative is if you're a large scale enterprise that deals with pabytes scale data our first initiative is how do we make sure that you find the right document given a certain query your agent sends
06:03you know set of queries to the corpus afterwards.
06:05Once you find those top five quer documents whatever it may be then we say okay you found the right document now how do you then find the right information within the document and this is where the work with the docysteine team has gotten really great where we've uh worked with them to build the neatron parse model to
06:20focus specifically on table extraction which is a really complicated technique um if you think about it the number of permutations of tables are quite vast when you think about nested tables merged cells merged columns merged rows whatever it may be and that can get really really complex and really hairy of a
06:36Uh so real quick as I mentioned before right our team is responsible for uh leading a lot of the leaderboards in the retrieval space.
06:42So Vidori V1, V2, V3, MTB, MMT.
06:46Um so our team knows how to build world-class retrieval models given a lot of leadership uh given a lot of leaderboard winnings that we've had in the last year or so and then of course
06:55as I mentioned before we open source everything right so we share the open source model weights the techniques and then we release with those blueprints and skills that agents can use then afterwards
07:06um so to touch a little bit on the actual model that we are working with with docuign was the neatron parse model so when you think VLM you generally think a multi-billion
07:14parameter model it's very heavy It's high latency.
07:17Um, this is a very small tiny C radio VLM.
07:21It's about 850 900 million parameter model.
07:24Uh, designed to kind of be that all-in-one package sort of model where you deploy it and instead of having small, let's say, YOLO X models that do table extraction or page element extraction, whatever it may be.
07:37This is a singleshot model that you can feed a document in and out comes the semantic formatting layouts the text uh the reading order uh the the preserved structure of the table etc.
07:47Um this can be served via the NVIDIA NIM or via the LLM as well too.
07:52Um and so it's a tiny small model that you can use.
07:54It's not a generator.
07:55It's more of an extractor at the end of the day.
07:59Uh so real quick as well too um we always want to make sure that we're building towards benchmarks that matter most to the enterprise space.
08:05So we want to make sure that both on the paro curve of accuracy versus performance, we're make sure that we're going to be releasing a world-class models to the ecosystem too.
08:14So you'll see here generally is just a very standard benchmark of table extraction.
08:17I believe this one was RD table bench and we compare some popular open source models here and then we compare how our neatron parse model does compared to that industry and we continue to kind of strive to improve this as time goes on.
08:30So that I think we believe we have a demo as well.
08:32Yeah,
08:34just press one.
08:35Yeah.
08:35Okay, there we go.
08:36Let me show you how easy it is to turn any agreement into structured usable data with agreement manager, which is a central repository of every agreement an organization has ever signed.
08:47Let's look at this.
08:49So when we look at the agreement manager view here, you know, we have an ability to see the entire list of agreements, but also go and upload a new agreement.
08:59So I'm uploading a new order form into agreement manager.
09:03As you can see, I can select from my computer, import from other places.
09:07The moment I select the agreement, it starts uploading and starts processing with AI.
09:14And just like that, you can see that the jobs engine has processed it.
09:18Let's take a closer look at this agreement.
09:21So when you go into the action, you can go and browse the file.
09:25Within seconds, agreement manager has extracted a rich set of metadata.
09:29Everything from key terms to commercial details are automatically structured, highlighted, and immediately you can jump to that section where the details are found.
09:40Builtin goes deeper.
09:42This is where the Nvidia's model comes in that it's extracted all the structured pricing data around this agreement.
09:48It goes in breaks down these complex t tables into order details.
09:55And as you can see, we can break it down.
09:58We can download all of this data.
10:00This is powered by the advanced parsing leveraging Nvidia's Neotron model, turning every even dense tables into something that is instantly usable.
10:09And of course, you can take this data with you.
10:12You can see when we've downloaded into CSV how we've structured all of it for your finance team, procurement team, even further analysis.
10:19All of it is also available through API.
10:22And that's how agreement manager has transformed agreements into actionable insights in seconds leveraging Nvidia.
10:30So I think you know what you saw there from um a demo perspective we've tried to shortened it.
10:35It's like we have a whole repository what you see a list we get customers which has thousand agreements to all the way millions of agreements within but the big piece is how do we understand and get that data
10:47that makes it very valuable to an end business user right a legal person a procurement person a salesperson who's doing a lot of the deals or even a leader right like a business unit the CTO goes and asks what did we do
10:59this is how we have making each of the things a lot more structured so we have our own proprietary agreement data model which we are structurizing each agreement but also at a whole organization
11:09level and leveraging a lot of the NVIDIA things we've been able to do a really good job especially with all of those tables like pricing SLAs's and then make that available
11:19and then we also have a like robust kind of search that is um on top of it so when we think about what have we learned right when you think from a neotron plus docyign
11:30we one of the biggest things for us we definitely have done lot of different models mod for different purposes.
11:36So a purpose-built model for the job you're trying to do is a big big part of how we've been thinking about and that's kind of where we've been able to accelerate bring things to market much faster.
11:47The second big piece around like the model efficiency.
11:50So for you know as Sean was talking about the number of parameters yes context and stuff matters in the you know in a different environment for different things.
11:58For us the lower kind of context basically also meant lower latency lower cost to deliver the scale that we are talking about last around the faster extraction.
12:09So um we we ran this against a lot of the other open source models when we think about how many tables can it extract per second Neotron was 20x faster
12:21which helps us when we're talking about the millions and billions of scale that we're kind of serving for all of our customers.
12:28So a lot of it is like having that smaller purpose-built things is the way for an enterprise as an organization to go and leverage and then serve that from an enduser perspective.
12:41Um and then what's next?
12:42So I'll let Sean talk through those.
12:45Yeah.
12:45So working with the docuign team uh has been awesome so far.
12:48Uh and we're going to continue to deepen that partnership as well over the next few months.
12:51So uh with them we started with the how do I extract as much possible information from a page and now we'll scale to how do I now find that page to begin with.
12:59Um so we'll start a little bit with the neatron retriever uh effort and then of course we'll talk a little bit about the NVIDIA agent toolkit with them over the next few months
13:08um and then actually start scaling into into more production scale agents then.
13:12Perfect.
13:13I think that's what we had.
13:14We have time for a couple questions.
13:17I went in the room.
13:22Okay.
13:23Someone there.
13:44So just to recap for everybody if you didn't hear it was the question is right like OCR is always a thorn in the whole process so are we thinking about letting the go of that and starting from agentic from the get-go
13:55I can talk from my perspective but so I think for us right like there are different use cases at different points in time many times if you are reactive you have a question and you're coming some of that can can
14:07work dynamically at a smaller scale.
14:09The question is the latency when I am quering at that scale of thousands I do need to have pre-processed have identified.
14:17So that's one.
14:18I think the second big part of the use case for us a lot of times businesses want to use this data to do a lot of downstream work.
14:25So an example is a procurement team.
14:27This is my pricing table.
14:28I want to put it into Koopa to make sure when I'm paying that works at that time there like you know the agent is kind of helping but I can't do that on a one document by document.
14:37That said, there is ways that we are compressing.
14:39That's kind of why Neotron worked for us is like how do you do it from a layout understanding just for that purpose but I would like let you add.
14:47Yeah, I think it depends on the use case a little bit.
14:49Um I think for this specific instance, right, you have pabytes of documents that you want to be queriable at some point, right?
14:56So you are heavy on the compute at the upfront side with all the OCR so you don't have to worry about it later on, right?
15:02I think there's some instances where people may upload a contract to begin with for Q&A and that's a very high that's a very low latency use case, right?
15:08So you have a high throughput versus low latency use case and in that scenario your different batch sizes, your concurrencies, your different techniques on how you process the document will be different and where you spend that compute in that cycle will be changing between the different use cases.
15:22Okay, one more there.
15:35We we do a lot of like more of what I call hybrid approach and a purpose-built for like the needs and the use cases.
15:42So from a table piece it does kind of you know do the whole layout along with extracting we still do OCR from a lot of other fields and metadata and a closet like all of the text kind of thing.
15:52So that I think we had a architecture where we have our pipeline going through two different routes for that.
15:58Um as a follow we have a blog out there how are we really solving this at scale across and if you look at that there's a lot of different
16:05peacemail modules and stuff together.
16:08Yeah one moreization
16:14so we this model is currently on FP16 but there are paths towards going on to FP8 and VFP4 in the next few months as well too.
16:22What about center?
16:23Are you using the 16 or
16:26we do use that and then we're also kind of using some of the older ones and that's the journey as a partnership is to kind of go tweak as you get more of the customers.
16:34Yeah.
16:34So for this there are many techniques on how to improve the performance side right so quantization right so we're trying to move everyone to blackwell right so that's why NVFP4
16:42is the big thing now um as well as multi-token generation for this it's a VLM architecture right so your encoder decoder techniques can definitely be further optimized so not right now this model just generates one token at a time you can do multi token generation of course too so there's plenty of performance things right now we're focusing
16:58on the accuracy side like are we adding value to the system and then from there we'll then push out that paro curve on the performance Are you guys using
17:13I believe they just deploy via directly recall when you do that.
17:29Oh, this is just an extraction.
17:31This is now a retrieval.
17:32Yeah.
17:33Yeah.
17:33Yeah, you're asking question.
17:35It's coming from that agreement data that we've kind of extracted and stored.
17:41So yeah, maybe I can chat with you offline and how like we our architecture kind of works fully as well.
17:49No, we're almost coming up on time there.
17:51Um, but I think that's kind of all we have.
17:53I'm happy to hang around uh in the back with more questions.
17:57Um, and good luck with a lot of your uh challenges with AI.
18:01So, thank you.
18:01Thank you.
18:03[applause]
18:18[music]