Name: Glean for the Legal Department
Uploaded: 2026-10-20T18:16:28.713Z
Duration: 59 min 56 s
Description: Glean for the Legal Department

Transcript for "Glean for the Legal Department": Hello, folks. Welcome to our Lien for legal webinar. My name is Michael Moore. I lead the Lien legal team here. I'm joined by my colleagues, Lauren and Amanda on the Lien team and Kelly on our marketing team. So I will start by giving some examples of how we use lean ourselves within our own legal department. And this is based on our own day to day workflows, how we use it for all kinds of different legal use cases. And the examples I'm going to show here are based, very much on things we act... We have built ourselves and things we've helped our customers to build. I will share my screen here. Nope. Sorry. Okay. So if you can see my screen here, what I'm showing here is the, clean for legal customer training. This is an environment that I built using lean itself. And I'm also gonna show some actual examples of using lean to to give use cases that are shown in this specific environment. This environment here was created using the Lean Tau tool, which is a tool that we are releasing in the very near future. And it allows you to work both locally on your machine, as well as also work on the, data that sits in your cloud within the standard lean environment. It brings all of the capabilities of lean and the capabilities of local work on your machine together, which can enable some very powerful use cases. So what I will show here is, first of all, legal work depends on context. Anything that we do needs context from across our business, be that from email or, chat such as Slack or Teams, be it from a contract repository or from a customer relationship repository such as Salesforce or HubSpot. All of that information is relevant to the day to day work that we do within clean and how we run our business across our our organization. This environment that I've created here is fully navigable navigable by users, and this is all synthetic data. It's built so users can walk through it. You can listen to an audio overview of the environment, and we will share the login for this as part of this call. If I go across here to search and chat, the... Any AI tool depends on context to give you clear and authoritative answers. What lean does is it pulls all of the information across all all your business. It knows you. It knows your role. It knows the kind of things that are relevant to you and how you operate. And lean is able to give you the information that is relevant to the question you're asking, relevant to your role, relevant to your business. The key advantage here is that when you run a prompt in lean, it understands the task, it understands the context of the prompt that you're trying to run, it understands the boundaries that is what data is permitted or not permitted as part of the of the task, and it uses that to give you the output. The net effect of this is you get a very tailored high context, relevant prompt, and response for you, which I find gives you a much, faster time to get to an answer than just using a generic AI, tool, which does not have the context of your business. As an example down here, within Lean, you can, for example, if we're looking at a matter context, these these are all a manufacturer examples using synthetic data. In this case, the question is, what should legal know about Meridian Analytics acquisition before deal sync? What Glean will do is go pull the data from sources, be the email or chat or drive or from CLM, assemble all of those together, and give you a deal brief that actually assembles this into a structured relevant output. This is all synthetic data, but it's a very good picture of how lean actually assembles the data together to give you the kind of output that you need for any any particular task. The nice thing is Lean can assemble all the data and all the context from all of the records that you have permissions to see to give you one comprehensive overview. There's an example... So this is this is a deal brief example within this this legal workspace. I had lean on this tab here, generate a hypothetical deal brief. So my prompt was was quite simple here. The prompt was give me a hypothetical deal brief drawing for the typical sources, Salesforce, Gmail, Drive, Slack, Gong calls, and provide an anonymized output. So what Lean has done is gone and generated... This is exactly the kind of brief that I would use for a normal deal that I'm working on. It would pull together a summary, pull together everything about the account, everything about the product, the the employees, all the relevant information on it. It would then pull contacts from deal calls, from the Gong calls, Slack, and so on to understand why the deal matters to a given customer. We would pull the sources that the information is drawn from. So... And then signal and the implication for the deal. Of course, this is all synthetic data. None of this is actual customer data. But on running this on real data, we'll give you all of this information together in one place that enables you to understand what you need in any deal that you're working on. I find this particularly helpful when, at the end of quarter, when there's tends to be a rush of deals coming through to be able to quickly assemble all of the deal and context information that I need to understand what... Who's the customer, what do they care about, what market are they in, what do they ship, why is the product relevant to them, who's who... Effectively, who are the people on on the customer team, what's their position, what, story or, engagement would work best for them. And then the standard MEDDPICs type sales qualification and going through all of the information relevant to the deal. So lean can assemble all of this for you from across all of your sources and give you a very clear comprehensive business case on the deal, what's required there, and how to progress it through the completion. So this this overall is a very useful summary of everything relevant to a deal, and it takes five to ten minutes to pull this for for real deal and build the complete plan out. It'll give you a closed plan. It'll give you all of the relevant information that you need to work through a deal. So that's that's one example pulled, just prior to this meeting from lean as a hypothetical example. Real use cases, hold the same format, and give you a very nice clear picture of what the deal is. So if I go back to the search and tap window here, the advantage is you have all of the information that you need. All of the sources are clearly outlined. So for example, for any email, I can click on it and inspect source. It'll pull out the content from the email. Similarly, if I need to pull something out of a deal memo that's sitting on Drive or SharePoint, it'll pull out the relevant information. You can click on it very easily and see where the information comes from and how it's relevant. And Lean will pull all of these things and stitch them together for you in a comprehensive overview. That makes it far more efficient for users to understand what the current posture of the deal is and understand all the relevant data points in these various threads that I'm highlighting here that go into making a a deal review or deal memo. The net of that is you can very quickly understand what's going on in any account and quickly spend your attention on the right things that matter in the account rather than focusing on things that are not relevant to the customer. So that... That's where this... I find this particularly helpful. If, next tab I'm going over to here is contract review. So as part of my role, I'm responsible for all of our contracts both on the sales side and on the procurement side, also all leases, and just anything else that that we are required and obligations we're required to deal with as a company. What I find lean is very helpful for is because it has all the context of the company, I can feed a contract, for example, a procurement contract into the lean. I can ask lean questions about the contract, and lean will pull together everything from the internal procurement channel, which typically that lives on Slack, any procurement emails that came in from internal users. It'll pull together emails from the vendor and themselves. It'll pull together all the communications that are relevant. And when I review the contract, then I'm able to review it with all of the relevant contacts. So I'm not just looking at the four corners of the paper, but also looking at all the business context and reality that lives around that contract. I find that immensely useful because I don't need to separately assemble all of the bits of relevant information. They will pull that together for me and make sure that I have all of that as part of the contract review. It means your contract review tends to be much more grounded in reality and the business case for the contract rather than just being tied to the words on the page. The nice thing is with clean also, you can review from the customer side and say, look, I'm a I'm a customer buying a product from a vendor. Review this contract from my perspective as the buyer. You can also say, tell the lean, look, review review it from the provider side. If the provider is selling to me, what are the things the provider sells about? So you can review a contract from the customer side or the vendor side. You can tell the lean, show me the terms where we are aligned on, things that we don't need to negotiate. So So for example, if I switch this button here, I can see things. Okay. The the green items here are aligned on. So those we don't need to spend time on. And I'm able to to take... See both perspectives on the contract very easily knowing the relevant context for the business. I can use lean to compare contracts against playbooks. So it's very easy to build a playbook in lean and then review the contract against that. So for example, if a playbook is running as a contract, in some cases, there's a mismatch here, I can have lean tell me where the mismatch occurs and potentially how to fix the mismatch. So if you ask lean to propose, changes to bring the contract into alignment with your playbook or bring it into an alignment with details that have been agreed on a deal call or a call recorded in Gong, for example. The team will be able to pull that information together from multiple sources, be they emails or drive or Gong calls or elsewhere, and suggest modifications to a contract to bring the words in the page into alignment with what was discussed on the calls or emails or Slack or other communication. I find that very effective because it it allows me to pull all relevant context in. So I then know how... Which are... To modify the contract, and clean pulls all that contacts in together for me. So that's that's how we use it for contract reviews. Personally, I find this very helpful. I have to look at a large volume of contracts. In many cases, some of them are for purchases of services. Some of them are leases. Some of them might be IP licensing or employment matters or a huge variety of areas. The nice thing is because Glean knows me and my role, it knows all the deals I'm working on, it knows all the legal matters I'm working on, it's able to review the contract with whatever context is required to provide the best and most concise, overview and advice on the topic. Of course, I always review. The human always reviews and understands the the output and signs off from the output. But what lean is very helpful for is saving you all the time that you typically spend having to gather all of the data points and try and assemble them and understand them prior to reviewing. So I can make my contractor use very efficient because lean pull all the context in together and understands it. So when I ask it to review a contract against the playbook, it'll compare the contract to the playbook, but also have all of the business context that is relevant to that. So another use case that I find particularly helpful are questionnaires. In many cases, if you're if you're selling an enterprise or b to b sales, customers will have a security questionnaire, a privacy questionnaire, often an AI governance questionnaire. If you are dealing with RFPs or RFIs, generally, there's large sets of questions that you need to answer as part of the RFP or RFI process. And in many cases, they can be very time consuming to assemble all of the answers together and to validate that the answers are correct and current. But one of the use cases I built for Dline, which I think works really well, is I can create a directory of known good answers. So for example, for infosec, I'll generate a directory of known good answers. And then I can point lean at that directory and say, lean, review this directory of good answers, review the questionnaire, extract all the questions from the questionnaire, compare them to the directive of no good answers, give me a list of all the questions with answers cited to source from the directory of good answers. What that does is it allows me to very quickly parse through a large questionnaire, pull out all the questions, pull out the most relevant answer to every question from known good sources, and then generate a response that lists out all the questions, all of the answers, and the citation of source for the answer to make it very easy to check quickly. And so I can, for example, I can ask Lien to point out an approved answer folder, in which case it'll it'll pull things just specifically from our trust center in this case, and, know that the content being pulled is is qualified relevant content from the trust center. Or I can say, Liam, just pull everything you can together. So I can take an all authorized sources. And in that case, it pulls a much broader set of sources here. It pulls across everything. Now if you're if you're pulling from all sources, then you'll need to review it with a different lens than if you're pulling only from a set of qualified answers. But I can I can tell lean pull only from the known good answer directory or pull more broadly? It'll compile an answer pack and it can tell me which ones, for example, are backed by sources in the known good directory, which ones need to be reviewed by a person, and which ones have citations attached. So Lean is able to tell me, here's all... Here are all the answers to your questionnaires. Here's what is pulled from the known good source. Here's the one... Here are ones you may need to review separately because they're not pulled from the known good source. So I find that immensely, helpful and saves a significant amount of time when reviewing these kind of questionnaires. What typically on some of these large questionnaires, you might spend, you know, two or three hours depending on the number of questions, sometimes more in going through and pulling all the answers together and validating them. I can typically turn a questionnaire like that around in fifteen minutes with lean, and then spend about twenty or thirty minutes reviewing the answers. And in somewhere between thirty and forty five minutes, have the whole thing completed, which is a saving of several hours over doing it, quote unquote, the old way, which is just looking up answers of our questionnaires and trying to cut and paste them or match them into the current questionnaire. So this works very well for high volume questionnaires, or if I's, or if piece, anything where you need to query set of questions, pull relevant and correct answers, and present them in the the format that the the questionnaire provider requires. This works very well. I'll move over here to discussing agents. So agents are something you can build within lean, and it's effective way to build a repeatable, task or set of tasks that can be repeated, that can be scheduled, and that can standardize the way you you address a particular, problem or thing you're trying to solve. I find agents very helpful because if, for example, you build a legal agent to do a particular task, you can then share that agent across your team, have everybody use the same agent, have them all use the same workflow, effectively the same set of thing... Of tasks so it's done repeatedly and consistently, and provides the output in a given format. The the agents are very helpful to repeat standardized tasks that you want to run regularly, and allow all the team members to run the same agents so they're doing it the same way every time. It allows consistency and repeatability. Some some example workflow agents that we have built on my team, one of them is a first pass, contract review agent. So it allows you to upload, for example, an MSA or an NDA or a DPA or whatever. It'll take the uploaded document. It'll review it against our playbook. It'll generate an output showing where the uploaded document is compliant or non compliant with the playbook, and how you would, potentially edit the uploaded document to bring it in line with the playbook. So it allows you to do a very efficient review on whatever contract comes in, run it against the playbook, and have a consistent output format. The nice thing is you can also add other contacts in. So typically, if you just wanna pull in and compare against the playbook and the deal context, that's fine. If you want to say, also bring in additional contacts from another source like a deal room, you could click, and add deal room context in and then expand the scope of the context that the agent, pulls from as part of the review process. Another agent relates to a questionnaire pack. This is the the example that I gave previously on questionnaires. You can actually build that into an agent. So rather than running a prompt manually, you can build an agent that will take any questionnaire. It'll point as a known good directory. It'll extract the questions from the questionnaire. It'll find the answers to the known good directory and give you an output in a consistent format. So this makes it easy to build an agent that points to certain known good sources. You share it out to your team or, for example, security team or sales team and have them run and know that they're using the agent that you build in the format that you build, pulling from the answers that you specified. So you have a consistent approach. It's very helpful for ensuring that people are doing things consistently following the process across the company, so you don't have everybody sort of trying their own version or their own flavor and and getting different results. Another agent that I use regularly is a weekly legal pulse. And what that does is it looks at all the work that the team is doing. It pulls items from our task tracker and contract database. It pulls things from our help legal internal channel. It pulls things from our legal at email alias. And it's able to bring in all the task activities that my team has worked on for the week, and then give me a consistent report in a consistent format every week. So this is very helpful because what it means is I have a a... Effectively a weekly overview of the team's work that I can review, I can share to execs, and I know what's going on across the team. It pulls in all of the sources, for example, our our contract database, our CRM records. I can add other things in. So, like, for example, I can add things from our email, things from our Slack chat. I can pull in things from drive, like, if you had a board meeting, we can pull in board minutes. If I want to look at, renewals, I can pull in things from the CRM renewal list. So I... Effectively, I can build an agent that allows me to pull in all these different sources and give me a consistent view of what's happening across the team. That's very helpful and saves a lot of time. Because prior at other companies, I would have had to assemble a weekly report myself from all the pieces, which is quite a time consuming thing to do. This weekly legal polls provides that report to me in an automated way, and that means I have a history of what the team worked on every week going back since I bought the agent. So it allows me to have a consistent view of what the team activities are over the course of time. Another agent that we use is a governance watch or just any legislative watch for that for that nature. I have one that monitors governance legislation across US states and the EU. So it attracts any new bills, any laws that have been that have been enacted, attracts any amendments. The key thing is here, it it runs and just pulls up and says, hey, there's been an update to such and such law. This is this is what is updated, and here's what you need to know. It it... Rather than me spend time watching many different laws and bills across multiple jurisdictions, I can set this up to track them for me and tell me when there's an update and just update me when something has changed. So it can run, for example, weekly, check to any updates on a weekly basis, and then give me a summary of what's required. I could, for example, tell it to look at all AI governance sources. I could say, look only at specific bills, depending on what I want to do, and it'll then track and tell me what I need to pay attention to over over the course of time as these things progress through the through the legislative process. Very useful if you are look... If you're following legislative developments. Some examples that are relevant are if you are a company with stores operating in 50 states across The US, you'll want to know what's going on in state law. You may also want to know what's going on in municipal law as it as it relates to your states or to your facilities within the states. And this allows you to keep track of everything quite easily without having people manually look these up on a weekly basis. Another agent that, one of my team members here built is an agent who will work a classification check. So this is actually a very neat agent that allows you to pull proposed contractor engagement, compare it against the California ABC test. You can look at the job description. You can look at the reporting structure. You can assess all the prongs, the evidence. And then based upon the agent, help classify the role as employee, consultant, or somewhere borderline, and then flag that for review with employment council. This is actually very helpful because our HR team can use this to review any, new engagements that they're proposing to bring on and classify quickly. So they they run the agent. They know typically what bucket this role falls into, and those when they need to speak to legal to have a a more detailed assessment put in place. Again, you could... The agents are are... You can build and do almost anything that you want to do as long as you have access to the information. Of course, anything in the employment scenario, you always want human approval. So you need to have a human review the activity and the human make the final decision. But this is very helpful for giving a consistent approach to your, for example, your HR team or your recruiting team so that they can run their own post... Job posts against the check prior to posting them publicly. Another, agent that we have built is a competitor claims research. So this is actually quite useful in that you can track them... Statements competitors make about my company, across the websites or press or public filings. You can log the claims. You can find the source when the claim was made, compare it against our product materials, compare it against any benchmark data, and also flag things that appear to be false or misleading or not supported by evidence. So overall, this is a very useful thing to run. You can pay attention to what your competitors are doing, what your competitors are saying about you or about their own products, and effectively keep track and update your own database of relevant things to your business and to your market. I run a similar one like this where I look at old old discussions on certain AI topics across the the broader industry. It runs on a daily basis. It pulls any new information together and assembles it into a Wiki for me. So I effectively have a Wiki of all of the relevant information in the comp... That is relevant to the competitive space that we're in that I can test against anytime. So it's a very useful way to track and gather information and store it. I'll move on to legal operations here. So the way we run our team, we try very hard to run our legal, like, business operation. And that it means spending time looking at the data, understanding the data, understanding the trends, and how does it actually impact our day to day work. In this case, the data sources that we care about typically come from our legal contract repository, that's our contract database, our task intake records, and also any business signals that we're getting from across our business. We wanna be able to see what tasks have come in, what the status of tasks is, the tasks that are waiting or stalled, and typically how long we we spend working on deals. It's very helpful to be able to see... To say, look, my team turned deals quickly. Here's the evidence to back it up. And sometimes if sales says, hey, a deal is stalled. We're able to pull out and say, yes, but it's not stalled on this side. It's sitting with the other... The counterpart is legal. Perhaps send them a thing to try and get them to hurry along. So knowing where the contract sits and how long it sits on each side is a very useful data point to be able to share. In this case, as an example, if I look at a weekly... And this is all synthetic data again. If I look at a weekly, deal flow, I can see, for example, that in the past week, 42 new matters came in, 36 matters changed, 36 matters closed this week, I can look at the median first response to the matter. I can look at, for example, things that are missing. Sometimes, item requests come in that don't come into the proper legal intake process, and those are ones that you have to identify and make sure that we capture them and place them in the legal intake process so they don't get missed. We don't want things falling to the cracks. So in this example here, it's very easy for me to see what new matters came in. I can just query lean and say, lean, tell me what new matters came in this week to the legal team. I can ask lean what matters closed this week. I can ask lean how long typically did it take for the matters closed. So it's able to to analyze and pull data from our contract database to see how long the typical tasks take and how we're spending time throughout the deal process. And this is all quite useful in terms of being able to analyze the team's activities and, understand where we're spending time with the team. If I want to look at deal coverage and where the deal sits, one of the common questions or common statements I hear from sales is the deal stuck with legal. But the question I have then is who's legal? Is it on our side? Is it on the counterparty side? Is the deal sitting with business waiting review? Perhaps nobody actually submitted a task for the deal, so it didn't get into our in... Our intake. So this is very useful to be able to pull out and say, you know, what deals we have expected this quarter? You know, which ones have a task attached to them. If they don't have a task, that's potentially a problem. I can see which deals are with the counterpart county party for review, and I can see deals that have not made it to our legal intake. So they may be things that come to us by... On Slack or otherwise, but have not actually been put into the legal task pipeline. So as an example, there are some sample records. Again, this is all synthetic data, but this would show, you know, for a certain deal legal review, our council needs to make a decision. Sometimes things are in business review. It's discussing pricing or or other or other services to go with the deal. They may be waiting on head of sales to make a decision on that. In other cases, deals are with the counterparty. We've sent over our our edits and red lines to the counterparty. We're waiting for them to respond. But it's... The nice thing is that Veeam can pull this information out of all the context that it's connected to. So you get a pretty good picture of what's going on in every deal, and you're not searching through a variety of different tools trying to figure that out. The other thing that's quite helpful is when I'm planning for budget or headcount. So every year we do an annual planning, which most companies do. I'm able to use lean to determine, like, how busy or how loaded are different sections of the team. Right? So, for example, again, this is synthetic data. If I have a commercial council team of four, I can see the amount of work that the team is doing. I can see, generally, that team is reasonably well balanced. They're doing okay. If we have, for example, one product or privacy council and they're spending all the time on questionnaires and vendor assessments, they may be overloaded. So I can I can see what sort of the workload is in a on a a given functional area and decide, do I need more resources in this area? If I'm looking, for example, a corporate or entity, I can see, do we have enough capacity for this particular role or task? How loaded is is the team supporting this? Can they take on any more capacity if they're under loaded right now? And, of course, for legal operations, I want to know how our invoices coming in? How are we tracking them? Are we are we validating the invoices against the task that were actually done, where the invoices sit in the payment process. All of that is is easy to pull in, and I use lean to analyze all of my invoices, understand where they are, understand where they are in payment, understand... Actually, critique the invoices for any any item that may not be consistent with our billing guidelines. So all of these are are use cases you can use glean for. The next one here actually is IP harvesting. This is one I actually really like. I I spend much of my career in the product and patent space, so this is one that I find particularly helpful. In most companies, the way you do a patent harvest is you set up meetings with your engineers or your software developers. You sit down with them and ask them, hey, Joe, hey, Mary, hey, Pat, tell me what new stuff you worked on. And just getting time with your engineers can be a challenge. Getting them to actually disclose can be a challenge. In many cases, they're hesitant to disclose things that they invented because they don't fully understand what's considered an invention. But all of the information is available for us to search because in most companies, the inventions themselves will have been typically engineering tickets or product design documents or in specifications on Drive or SharePoint. They can sit in email threads or in chat discussions on Teams or Slack. All this information is there, and as long as you have visibility into it, you can then assemble all that record into a complete picture. What this means is it makes it very easy for in house IP counsel to assemble all of the information relating to products and features that your team are working on, and then assemble a nice tight, invention disclosure or list of inventions on that that you can use for patent harvesting or just even to understand the invention coverage and what's... The r and d team is doing across the company. So this one I really like. Again, this is a synthetic example. But for example, I can... I'll show you a real one in just a little while. I can ask lean to go pull together all the information on, you know, for example, a field telemetry prototype. It can then... Lean can then pull the information, understand what is the problem, understand what approach is being used to solve that. Are there any open questions for patent counsel on this one. And it'll pull all of that information together, and then I can assemble that into a into a document from the sources and queue it up for patent review. It'll assemble a a complete package that I can then review internally or send to asset council for review. Another another example concept here is a predictive maintenance concept. In this case, it pulls the information together from a variety of systems, be they from drive or product wikis or any other sources here. And I can assemble all the information together and queue it up for pattern review. The nice thing is all of the information that assembles is cited to source. That makes it very easy to check quality underlying documents that this was drawn from. For example, if it's a product Wiki, I can pop open the product Wiki. I can go to the actual section of the Wiki and understand, here's the source. Here's here's the the data that we're using to assemble the patent, or invention disclosure from. I actually find this very powerful. There's one example I have over here. So this is an anonymized example here. This is this is pulling features that Lean has shipped. These these are things that are, shipped and public. What I... What Lean has done is I asked Lean to research Jira, Confluence, Drive, Slack, and pull together an anonymized roadmap of all features Lean released over the past six months. I'm saying it will be shared publicly so I'm not... Anonymize the names of any individuals. What Dean has done is gone and pulled all of our release notes from from public information and given me a complete list of everything that is shipped. So we see April. I get down further. There's a big long list as you can see. Then May, things we shipped in May. So we've shipped an awful lot. Then in June, July. So this is a complete list of all the things that we shipped. There's August and so on. Now that makes it from a... From perspective of product council or IP council, that's really helpful. It allows me to go through and see everything that we shipped and, when we ship them and from the purpose... From the perspective of making sure we capture any IP relating to them, it... Lean makes that very easy to do. I can even go and have Lean compare the ideas I have versus what's already in the portfolio. So I'm gonna step over to the next, tab here. This one I find very helpful also. So if a new idea comes in or I get a list of ideas, I wanna check, has that already been patented? Or do we have an existing invention or patent filing on that? Now, in larger companies, that can be actually a complex task. Because if you have a large number of of existing patent filings and issued patents, knowing if a new item that came in is relevant to something that was previously filed can be a challenge. You have to really understand your portfolio in-depth to do that. Most IP tools don't make it very easy to search that kind of question. So what Bling can do as long as it has access to the underlying inventions and can read the files, then, for example, I have a candidate invention here. I wanna check, was this, something that... Or here... Here's, for example, the sort... The source again, synthetic data of that invention here. I wanna check it against the portfolio. So I can run it... Run a prompt which says, lean, review this invention, look at my directory of prior file patents, tell me if this invention has been previously filed based upon the technical content in the disclosure and the technical content of prior file patents. In this case, they'll get, an example back saying, yep, certain things, how... Were covered and certain things have not been covered. So it... Glean will be able to tell me, these are ideas from your list of ideas that you already have a patent filing on. These are ones that you don't appear to have a patent filing on, then I know which ones I need to focus my intention on. That's very, very helpful when you're going to large portfolios and trying to determine what things do I need to get coverage on that were not previously covered in the portfolio. The other use case I find very effective is an opportunity scan, which is where should I file new things on? What should... What signals should I pay attention to across the market in terms of understanding areas that would be interesting to develop new IP on and why? And understand not not just, hey, it's cool, but there's market signal behind us that actually makes sense to file in this case. There's a business reason to do so. So for example, I can look at my my own company's road map. Like, so for example, I can select things such as my own company's road map, pick that as one of the features I look at. I can look at the market. I can look at competitor product announcements, any new features. I can see certain better patent filings or publications. I can even look at videos or the press releases or other other comments by competitors. And I can use lean to assemble all of that together and get a a briefing or a status of the market in a given tech area. So I can look at my own company's road map. I can look at market signal out there. I can even get call notes from customers. So for example, often many sources of good ideas come from customer discussions where customer would say, hey, I'd love if you had such and such a feature. And those can actually be very valuable in terms of feeding into the product road map and understanding what things we want to build. The nice thing with lean is I can gather all that information together and then use that to assemble here are areas we should consider filing up. And I u... Use this regularly now and gather competitive signal, market signal, and assemble it on a daily basis, and then determine what areas should I consider filing more IP on and why... What's the business reason behind it, what's the competitive reason, what are customers set. So So it's a very useful way to assemble all the information together and then use that as as a decision tree in where you should file new applications going forward. I'll go across here further. If I go to litigation, there's some very good litigation use cases also. So for example, when I run litigation matters, what I find very helpful is I can put everything into a single folder and then review across the entire folder very quickly. I use the tool, LeanTow, to do that. We did run locally in my laptop and I can point it out litigation folder and just understand everything. Even if there's many, many gigabytes of data in there, Leantel will very quickly go through. It can pull out answers to any question I care to throw at it. It'll cite the source so I can pull out the underlying documents, be they emails or or invoices or, filings or anything else that are in there. And I can very quickly pull out matters in a way that speed up the litigation process immensely and save an awful lot of outside counsel time and therefore save money on outside counsel. So this is particularly relevant if you're managing litigation in house and you want to be as efficient as possible with your outside counsel's time and spend by doing a lot of the digging yourself through the the document folders that are available. I find it very, very efficient to do this. So for example, this is again a synthetic example. If I want to point at a matter folder, I might have 70,000 or 71,000 items in the folder. I can look for certain things that will post communicated about pricing changes before a certain March amendments in the dispute. So lean will pull up the relevant candidates and responsive candidates and documents from the folder. If I ask, well, another question, well, what do the counterparty commit and deliberate? I can put... Run that prompt. Lean will again run through the entire matter folder, pull out the response of candidates, and pull out key documents. I can also ask, hey. Did we have any similar dispute, issues relating to this? It can run through all prior matters as long as it has access to the documents and pull out any prior matters that are relevant to the current one. So I actually find this very, very effective. When I'm managing litigation matters, I throw all of my relevant documents into one folder. This this might... Maybe many tens of gigabytes of data. I point leaned out folder and just ask the questions I want. It'll sort through the entire stack of documents very quickly. Because it runs locally, it's extremely fast. And I can pull out answers to virtually any question that I care about on litigation matter, and I'm not spending outside counsel time to do so. I can do it myself. It's very, very quick and a very efficient way of handling it. Similar to that, there are compliance use cases that actually work very well also. So So in many cases, if you are a company that resells through channel or partners or marketplaces, if you're doing sponsorship commit... Commitments or marketing, funding commitments, you want to be able to understand how how the... How those those deals are flowing. Are the partners and channel following the guidelines you're giving them? Are they are they following your, any compliance terms you're passing through or passing through your own terms of service? Are they handling pricing or discounting correctly? And how... If you're providing with marketing funding for market development, are they spending it correctly and appropriately? Normally, that would be a huge stack of documents scattered across multiple systems, which is actually quite difficult to audit. Typically, you might do an annual audit or a quarterly audit and pick some examples and test against them. The nice thing is with the tool like Lean, if you have access to the online documents and you can point to those documents, then you can set up an agent or set up a repeatable process to run a quite detailed scan on a regular basis. And this is very helpful if you have... If you're running resellers or channel partners, to be able to look at, for example, all of the channel deals that were sold in a given, region or a given country. Analyze, for example, do the deals sold in this country look like other deals sold in a different country? Are they generally consistent or are they different? If so, why? So lean will let... If you can query the data, lean will let you analyze quite effectively how your channel program may be, consistent or not consistent by geography or by time period or anything else. It allows you to dig quite deep in and look for anomalies, look for things that that, for example, do not comply with your, channels, specifications and process, and quickly determine areas that you need to pay attention to. It allows you to find needles in a haystack quite quickly, so then you can spend the human time and attention focusing on the needles instead of having to sort through the whole stack. So this is a very effective way as long as you have the agreements visible and available to the lien. It can read them, it can understand them, and then you can ask whatever query... Questions or queries you want against it. And even build agents that then repeatedly check the same thing on a consistent basis. For example, monthly or quarterly as you choose to do so. We also use lean heavily for corporate matters. For example, entity work, board work, governance. The nice thing is these... There tends to be a very heavy set of documents, particularly when you're running multiple multiple entities globally. And tracking all of these and staying consistent with them can be challenging if your if your documents are scattered across multiple directories or tools. The nice thing is Lean sees all the documents. And typically, the way we structure them is we have, an entry folder with a subfolder underneath for every country we have an entry in. Each of those subfolders gets the relevant documents, notes, and emails for a given entry. And then then Lean can walk those directories and pull out, like, what's the status of a of a given entity in a given country. Do we need a new entity? For example, if I if I want to look at rolling out a new entity in a certain country or look at the status of entities in a given country. And this is this is... Again, this synthetic data, but this is reflective of the use case we actually have. Tell me what's going on in country a. Our payroll, banking, and tax set up. Tell me what's going on in country b. Like, when when do we plan to go live? What's going on in country c? Well, we're early on. The business case is approved. We've incorporated it, but we we need to do all the finance and tax and banking and so on. We're thinking about going into country d. In that case, it's on hold, pending resources review. So it's very easy for me to ask these kind of questions and get a very clear picture across my entity stack and corporate stack of all of the things that are going on. Normally, this would be a quite time consuming process. As long as I have access to the data, I can very easily ask any questions about any entity in any country and know what's going on there. I can also pull in information from the internal company directory, say, how many employees are in country a? How many contractors are in country b? And Glean will pull that information from the employee directory, combine it with the entity information for the entity folder, and give me a very clear picture of what's going on in any given, country. I can also... If I'm if I'm approaching, for example, a threshold number where I wanna form an entity in a con... In a new country, I can pay attention and run a a... An agent or otherwise check periodically to see how many employees do we have in this country now, or how many contractors we have in this country now. At what point do we need to actually form an entity in the country? And I can also pull in information from Salesforce or CRM and see how many deals we have pending in the country. What's the the dollar value of all those deals? What's the headcount for the individuals? And then I can start looking at, well, is there an ROI in adding additional sales individuals in that country? It... Does the market justify it? Are the opportunities big enough to do so? And it's very easy to pull all of that together within lean. So you're pulling from the corporate and entity information, you're pulling from the employment information, you're pulling from Salesforce for the opportunity information in the country. You can assemble a quite clear picture of what's going on in a given directory. Sorry. In a given country and why would it benefit us to expand there or not and for what reason. So another key area here is in order to actually run a tool like lean or any other tool and rely on the answers, you need to know that you have the right data sources connect them. So the the actual... Having the right data connected is a key part to getting full value of the tool. Typically, when I'm looking at a customer as a new customer onboarding, the question I'll ask is, are you connecting all your key data sources to the tool? So for example, if you're using it for GTM, are you connecting in your CRM? For example, Salesforce or HubSpot? Are you connecting in the channel that your sales team communicates in? Be it email or Slack or Teams. Are you including the sources where they store data? For example, Drive or SharePoint or other systems. Are you pulling in your customer call records? For example, Gong or Zoom or Gemini or whatever tools you store your calls in. You need to make sure all the right tools are connected in so lean has access to the the data that you need to answer a given question. Then you also wanna look and see, are we covering the right time period? So, for example, if different tools change over time, for example, a company brought in Gong six months ago but used a different tool prior to that, you wanna make sure that you... The data period that is covered is appropriate to the problem you're trying to solve. So if somebody brought in a new tool recently, but it only has a short, period of data, then you may wanna consider adding... Connecting in or linking to the data from the prior tools so you have a greater period of coverage for historical reasons. You also want to know, like, where are your sources of record? Similar to where I I looked earlier at having a known good directory of an... Directory of answers to answer infosec type questions. Know where your your source of record is. If you have a contract database, build a point and connect lean to the contract database, so you know you have the latest version of agreements. You have the executed copies in place. We have connectors of Ironclad and some other other databases. I think the the key thing here is making sure at least what system you use is connected and visible to lean. One of the the real real key things with lean is that it is very strongly permissioned. That means lean inherits the permissions of the individual user running the query. In my role, I tend to have fairly broad permissions to see across a wide variety of things. Somebody in a different group might have a different set of permissions. If we both run the same query, we'll get an answer that is tied to our permissions that's in the online documents. So that's very important because if you're running a privileged legal analysis, we wanna make sure that only the legal team and perhaps certain key business stakeholders can see it, not the whole company. So Lien is strongly permissioned As long as you set your permissions correctly in your source documents and systems, will follow those carefully and make sure that any answers that are presented are permissioned according to the person asking the question. In addition, if you're... If you are running, legal research, what I find very effective is I will always check the source of the research material that I'm using before using it to answer questions. So what I found quite helpful is many of the law firms publish annual guides on privacy or corporate or taxation or employment or so on in different jurisdictions around the world. Those are generally authoritative and high value legal content. I will take all of those. I'll I'll drop them to a set of directories, and then I can point at those directories and say, hey, Darlene, review this directory of Singaporean law that I received from law firms. Review this... The the Singaporean law or do any government guidance from Singapore, which I'll typically find and drop into those directories too. And then based upon that, analyze this employment scenario in Singapore. Here's the employment agreement of the individual. Here's the statement of facts. Here's what's going on. And I'll be able to... Because I've I've I've linked Lean to authority of legal content and sources, be able to use those to to construct a very, detailed and well thought out answer to whatever the scenario happens to be. So I find that very effective. If you if you tell Lean, go search publications by top tier law firms, pull down a set of articles based upon privacy in various jurisdictions, you... Lean will go fetch those articles back from known good sources, and you can store them and then use them in the future for future almost effectively as a a binder of authoritative sources that you use going forward. If you want to do a pilot, what I say generally when... If you wanna pilot lean, we encourage you to do that. We would love to to have you learn more and, engage with us. If you're picking a pilot, I would say generally pick a, couple of workflows. Pick things that that are well... Clear, well rounded tasks that you want to understand. So for example, contract review or questionnaire for privacy and security. Right? I can pick those items. I can then say... I can pick on one of the the success measures. Do I wanna look at the source coverage or the quality of the edits? Do I wanna wanna look at time response? And then, effectively, generate a pilot outline here that shows out these are the things that I want to look at, commercial and private security, and here's the timeline for the phases, and here's the things we check again. So, if you're interested in learning more, we'd love to speak with you. The key thing is getting a pilot set up. Let's actually test and try lean. I built this, environment here, and we can provide the password in in the chat to folks to actually log in and walk through this environment. It is narrated. There's there's an audio talk through with the button that's in the chapter on every one of the pages. So if you wanna get that sort of a feel of the use case that lean can be used to solve, you can actually go through, walk through this environment. You can listen to the chapters and get a better idea of the use cases you can use the info. Everything in here are use cases we have built ourselves. We use on our own legal team. So these are all real, they work, and they're very effective. If you'd like to learn more about them, then please do reach out to us. We'll be very happy to to speak with you about that. So we would encourage you to keep keep learning here. There's a link to the Dlean for legal page here. It's on, the lean.com web page. There's a blog on AI for legal teams, also on the Dlean web page. And if you wanna speak to us, please reach out to legal of gleam dot com. We're very happy to talk to you. I do a lot of customer facing training and, customer engagement. I've spoken with well over a 100 customers at this point. I've got a pretty good idea of challenges that that, legal teams are seeing in terms of adopting the various legal tools out there. And in many cases, there are a lot of good vertical point tools, but they don't pull in the full context of your business. The reality of our of our legal support lives in the business. It lives in email and chat or Slack or Teams or SharePoint or Salesforce and all of those other tools that the actual underlying source data sits in. Lean sees all those very effectively. So what I find is we can use Lean to run very efficiently, and very effectively as a team. And also, we we... Because we run efficiently and well, we can get things turned quickly, and we're not working, you know, morning to night. We can run a much more balanced... Much much, better work life balance because we're able to tackle things efficiently and quickly. There tends to be fewer fire drills. And even in the quarter, we're able to handle the the full workflow that comes in during that time. So I'd encourage everybody here, take a look at the the department's legal web page, take a look at the blog, reach out to us at legal@clean.com, and, we'll share the password and log on to this demo here so people can log in and test walk through it yourselves. Think of all these use cases. But I would ask for attendees here if any of this resonates, if there are particular use cases you're interested in, please email us. Send an email to legal@lean.com. We'd love to hear from you. If there are features that you think are missing, again, we'd love to hear from you. If there are use cases that you built that you found to be quite successful, we'd love to hear from you. So please consider this as an open forum to send us any feedback, questions, any commentary at all. We'd love to hear it from you, and we will we will assemble all of this and, create additional content, additional modules to build out this presentation based upon the feedback we received from the audience. So I'll turn to my my colleagues here. Any questions come in on the chat that you'd like me to take a look at? Yes. Thank you, Michael. This is great. The chat and q and a has been, busy. A lot of questions are like what prompt you used to create this artifact, as well as the agents you used for AI governance. Just overall, I think folks wanna just learn more about how you do it. Like, what are some of the prompt you used? To to build this artifact? Yeah. It was actually really easy. And and it's... The the... What I what I asked, I used the tool, Lean Tau, which is as our... Effectively our desktop local operating version of Lean. And I said, Lean, look at all the use cases that I've discussed with customers. Look through all my call notes. Look through all my my past customers' customers. Look at the use cases actually built within Lean itself, and then give me a list of all of these use cases. So that Lean went through all of my emails, all of my calls, all of my everything, and actually built out a list of these use cases. And so so then I said, okay. Now what I asked them how to do was build me an interactive environment that I can host on Vercel that people can click through and listen to, and give me a web page for each of these categories. And then I asked it to just build it build it out with the lean theme and colors, build out the content that's relevant to the actual examples that I do, which lean is aware of, and build out the content for me. So, I mean, lean went and built the whole thing. I was able to model it and run it locally on on my machine. And then you connect it to 11 labs to generate voice for the for the overviews. So there's an audio track for each of these these pages. And lean is able to build all of the content. It was able to create the audio. It was able to polish the the process, throughout. And it took me... The whole presentation I showed today took me maybe four hours to do. Now that's pretty impressive for a fairly detailed interactive website to build. The nice thing with lean is I used, two models. I used OpenAI Luna, which is a very inexpensive model to prototype it. And then I used Kimi, which is a frontier model, but a very reasonably priced open weight frontier model to actually build the final results you saw here. So the the prototyping was done very quickly and inexpensively. The final build and polish was done with the frontier model. The total time to build was about four hours. The entire cost of building this, in terms of tokens used, was about $35. So that's very, very cost efficient in terms of how I was able to build this. And that's one of the advantages of using lean. You get to pick whatever model you want. You... Lean will automatically pick the most cost effective model. So you can prototype and do a lot of background work using an inexpensive, lighter model. And then when you want to do a final fit and polish, you can use a Frontier model at a very, very reasonable price to actually put the final polish and assemble it. So the... It was... That's how I built the entire site on the on the the process here. The nice thing is anybody using lean can build us. Right? This is this is easy to build. Many of our salespeople have built environments like this. It may... If you want to demonstrate or build a training, for example, internal training platform for your own company and team, Lean can help you build out that training platform, make it interactive, make it fun, put some audio on it, and may just generally make it much more engaging and much more interesting platform to teach people, even if you're doing internal compliance training. Or, hey, here's how to interact with legal rather than just put an FAQ out there. If you actually make an interactive demo like this that people can walk through and click through, I think they engage a lot more than just reading, you know, plain black and white FAQ. Yeah. And, also, I think, it's not like a one shot, one prompt. You get this beautiful artifact. You can keep iterating and workshop. Yeah. link. You can share it with other... Your colleagues. They can provide feedback. You can just keep tweaking it. It'll give you a v one, and you can keep working off it. So. just takes the pressure off, in terms of playing. with me. And this... And by the... I I built this. I iterated on it a few times. I used voice input. I didn't type any of this, by the way. The whole thing was done using Lean voice input. So I used Lean voice. It provided all of the... I just... I looked... As I as I reviewed every page, I would have my voice button passed. I'd say, hey. Tweak this, tweak that, move this thing here, move that thing there. And Dean would just track all of my commentary in lean tau. So I'd... I would open a... As I'm iterating through the page, just sort of read out my my feedback as I step through every page. So on page three, do this. On page four, do that. And just give the feedback to Lien. Lien would do another spin on it. Five minutes later, we'll come back with, an updated page based upon all my feedback. So I took maybe two or three rounds of iteration doing that. So it is very, very quick to actually build out and iterate and polish the the artifacts. And then I I I told Dean, I wanna host this on Vercel. Give me a package to upload to Vercel. So Dean just give me the whole zip file and push that to GitHub. Yeah. Vercel pulled it from GitHub. The setup was easy. Very easy to do. Yeah. And this is actually part of our monthly series where we start with, like, a Glean one zero one on... Explain what Glean can do and then take you through the journey on, like, very use case specific, persona specific content like this one, Glean for Legal. And next week, we have a Glean two zero one session about agents and skills building. So you can learn from one of our, customer success managers about how to actually hands on be building agents and skills that Michael showed. And then if you missed the Glean one zero one session early this month, in two weeks, we'll have, kickoff our November series, and we'll do it all over again. You can register for the Glean one zero one, to also get, like, a overview. So I posted those links in the chat as well. Cool. So, hopefully, we addressed all the questions. Is there any other questions we didn't get to answer? Any other feedback, comments from folks? I'll put I'll put my LinkedIn profile here in the chat. Feel free to connect with you folks. I I speak and publish quite a bit on these topics. I have a number of published articles up there that are on my LinkedIn, you're very happy to to review. I'm happy to send you copies as well if needs be. I would just say, keep keep the communication up. Like, reach out to us. You have the legal@clean.com email. I'll put it in here. Send us any feedback. Send it to you if thought this is helpful, not helpful, anything you'd like to see different, any feature requests. Just... I'd love to hear your feedback in terms of how you use the tool. Things you found it successful with, we'd love to hear that. Things you didn't find it successful with, we'd love to hear that also. Just tell... Give us as much feedback as you possibly can, and we'll do everything we can to make the tool better and provide more training like this really directed towards this audience by our own in house practitioners ourselves. Yeah. Thank you everyone for joining. Stay till the very end, and thank you, Michael, for the great presentation. Thank you, Lauren and Amanda and Honey, for answering the chat. We'll post down the main link. Once it's ready, you'll get an email. So please, forward to your colleagues, who missed it. Thanks, everyone. Thank you, ma'am. Thank you all. Bye bye.