How Anthropic teams use Claude

0:04 Hey everyone, welcome to our breakout session today where we answer one of the questions my team and I get asked all of the time. How does Anthropic use Cloud? My name is Eleanor Dorfman. You might have seen me earlier. I'm the head of the industry sales team at Anthropic based here in New York. And I'm really excited to kick off this session today where we're going to hear from four leaders from our legal, product, finance, and sales team about how we use Plaude, co-work, and plugins every day in our day-to-day work. Each of our speakers are going to walk you through real use cases, and then we'll do a quick demo that really brings them to life. To kick us off, I'm really excited to invite Lindsay Schwitzer up who leads analytics for a finance and strategy team. She's going to talk about how this team uses dashboards with co-work to run their daily lives. Please come up, Lindsay.

0:58 Thanks, Eleanor. Hi, everyone. Thanks for joining us today. I'm Lindsay Schwitzer, analytics lead on finance and strategy here at Anthropic. Today, I'm going to walk you through some ways that our finance and strategy team uses Claude and Co-work. Everything we do starts with the model context protocols or the MCPs. They're the foundation of how we work. You can think of MCP servers as the connectors that let Claude actually do things in your environment. So in finance, we have four core connectors that are wired up to our cloud. The first one is our data warehouse connector. This is the source of truth for all of our financial data.

1:39 We have a messaging and collaboration platform for alerts, communication, and pulling context from all of those channels. We have a project management tool for task management and project tracking. And we have our file and knowledge store for internal knowledge to pull for Claude. The key thing here is that these aren't just readonly connections. The MCP gives Claude two-way access to query, post, and update across all three systems just from a single conversation with Claude.

2:09 So while the MCP connectors give Claude the ability to talk to our systems, skills tell Claude exactly what to do once it gets there. The finance skills are built on top of the data warehouse MCP and they tell Claude what tables to query, what data to compare, how to format outputs, and then the MCP executes that skill on our live data. So a couple of the skills that we've built on our finance team here are shown up here. We have our financial statement skill. This one generates income statements, balance sheets, cash flow statements, and period over period flex analysis, which has saved our teams loads of time at month and quarter end.

2:50 We have our insight agent and our variance analysis skills. These ones encodes institutional knowledge about what database tables map to which key metrics. They have golden answers that are validated by data teams and they include instructions on how to find the why behind our numbers. They give our finance team a unified interface for data analysis in cloud code. Lastly, my favorite one, our Dashmaker skill. This one builds on the first three skills, enables anybody on our team, whether they're technical or not, to create beautiful interactive React dashboards with confidence that Claude is pulling all of the underlying data correctly.

3:32 The layering here really matters. So skills encode our institutional knowledge, the how do we do things here at anthropic part, and then the MCPs provide that live connection to data. So using them together, you get something super powerful. Let me walk you through a couple of concrete examples of what this looks like in practice. Skills and MCP is not just running queries. It's really synthesizing all of our data into a story that helps us better run the business. First is our alerts channel.

4:06 We pipe anomaly detection and trend analysis into a channel called ant stats alerts. Claude pulls context from multiple data tables, dashboards, correlates all of this data against historical patterns, and surfaces what we actually need to focus on. These alerts are not just notifications. They're contextualized insights. So instead of getting something like revenue was down 3%, you get revenue was down 3% due to a drop in enterprise renewals which broke from the trailing six-month trend. This has majorly shifted us from being reactive to proactive about trends emerging in our business. The second way that we are using skills and MCPs is for our cohort dashboard. Skills can execute scripts which means we can generate consistent and reproducible outputs.

4:56 Our cohort dashboard is a great example of how we're doing this. Cohorting requires really complex and layered queries. We're joining revenue data across time periods, bucketing by signup date, calculating retention and expansion by segment. It's the kind of SQL that's really tedious to write and easy to get wrong. Now, we're using Claude to help us define several methodologies. We've taught Claude the underlying data model through our other skills so that it can write these queries reliably and consistently. And from there we just iterate with Claude on the front end of what we want this dashboard to look like. This type of analysis would have taken us months before. And one financial analyst on our team built this in the course of a day.

5:42 In finance, it's really important that we trust the numbers that are being pulled and presented and that we trust that they're going to be pulled the same way every time. Now that we have that trust, we have access to near immediate insights from the analysis ideas to outcome is shortened from months to literally minutes and hours. And lastly, the development model here. Claude is actually writing all of these skills for us. Nobody hands codes these skills. They're markdown files. There's no infrastructure, no build steps, no technical requirements. You describe what you want. Claude writes the skill.

6:21 You test it, iterate, and you're done. The clo co-work plug-in management meta plugin exists specifically for this. With Claude, our development velocity has gone from weeks to spec and build to hours to prototype and ship. our cohort dashboard, our department expense views, our peer-to-peer metrics. All of these are built by describing what we needed and letting Claude scaffold the queries and the React components on its own. There's no front-end engineer bottleneck. There's no design review for internal tools. The last piece I want to mention that's really important is the sharability of these skills. So, we have a companywide data warehouse specific skill that teaches Claude how to query all of our tables. Anyone at Anthropic can install it and immediately ask questions against our data without needing to know schemas, table structures, or really even how to write SQL or code. It's our institutional knowledge packaged and distributed for everyone's use.

7:23 Now, I'm going to turn it over to Alain to do a live demo that showcases our cohort dashboard. >> Thanks, Lindsay. So Scott earlier in the keynote talked about how skills are kind and plugins are designed to be a starting point and this is what our finance team basically took to heart. We have that Dashmaker skill that Lindsay talked about, but what they were able to do is actually build on top of that and build a much more comprehensive Dashmaker skill that takes in more information and is much more customized to the work that they need. So this cohort dashboard is a great example of how these skills get built. simply a markdown file. As Lindsay mentioned, this is all designed to be really easy to work with, even if you're non-technical. And of course, having Claude work with you makes this even easier. This cohort dashboard is much more comprehensive and really supercharges what that the Dashmaker skill that we're releasing today publicly has to offer. And here's what this looks like in real life. So, this is this cohort dashboard, which breaks things down into into different periods. You can see ARR by RCT cohort.

8:26 Lindsay, based on kind of this dashboard versus what sort of the out-of-the-box one comes with, what are some of the advantages you get with something that's much more comprehensive here? >> Yeah, absolutely. We can tailor this one specific to the exact analysis we need. We can filter it to just a specific product. So, we can really get granular about which cohorts are contributing to our unit economics the best. >> Cool. Thanks so much. >> Thank you.

8:52 I'm now gonna pass it off to Mark Pike from our legal team. Mark will tell you he doesn't know how to code, but he helped build our legal plugin in an afternoon. Mark, take it away. >> Thank you so much. >> Hi everyone. I'm Mark Pike and I'm on the legal team here at Anthropic. Legal teams are information dense, high stakes, and before clawed, chronically bottlenecked. Our job is to review things carefully before they move forward.

9:23 But carefully often means slow and slow means that people start to see you as the department that holds things up. That's the dynamic we wanted to change. Now, before Claude, I had a backlog of work I'd push to end of day. Important but tedious tasks that didn't use the best parts of my brain. Repetitive support tickets. Marketing reviews that start to blend together. Quick Slack pings that really pile up. necessary work but a grind. So that was the starting point. Not how do we use AI but how do we stop being a bottleneck.

9:59 I don't know how to code but I helped build our legal plugin in an afternoon. We started by pointing Claude at everything the legal team actually does. The memos, the risk frameworks, the policy documents, the way we think through problems at Anthropic. Claw distilled all of that into a plugin. Not one that does generic legal tasks, one that does ours. And then something delightful happened next. I ran the product launch brief for the legal plugin inside of the legal plugin. It was so recursive and so meta.

10:31 It flagged real issues based on our own frameworks and things that I've told Claude that I care about previously. The plugin reviewed its own launch and the launch was successful. And that's when we knew it was working. Now, I'd love to give you three concrete examples of what this looks like in practice. First one, regulatory monitoring. My colleague Pamela, shout out Pamela, she tracks AI regulatory developments across dozens of jurisdictions globally. New laws, proposed rules, enforcement actions. The volume is really unmanageable for one person without help.

11:07 Claude now monitors and synthesizes regulatory updates continuously and surfaces what's relevant to Anthropics business. Pamela reads what matters instead of reading everything. And when you're a company building AI and global AI regulation is moving at this pace, that's not a nice to have. It's it's essential. The second one, internal communications. Every two weeks, our department sends a legal update to our executives. What the team's working on, what matters, what people should know. That used to be a significant manual synthesis job. Now, Claude prepares the first draft, pulling from our work logs, translating legal work into language the rest of the company understands. We still review and edit. The update still sounds like us, but it takes a fraction of the time. Now we can focus on the important work instead of spending our time writing updates about the work.

12:04 The third one I recently analyzed 742 Jira tickets which is an oddly specific number so you know it's real. we wanted to understand patterns in how legal requests were coming into our team. That kind of analysis structured systematic pattern matching across hundreds of data points would have taken weeks manually. Claude did it in a single co-work conversation. The output shared or rather it shaped how we re restructured our intake process entirely and identified areas for automation. If you're struggling to think of where to even get started with how Claude can help your department, I'd suggest doing something similar. Have Claude look at your department's recent busy work and ask it for suggestions. This really solves that cold start problem.

12:53 So what is the lean? Sorry. What is the legal plugin? Technically, what's under the hood? The legal plugin is open on GitHub. It's markdown files, prompts, system instructions, configurations. It's not a legal specific frontier model. It's just clawed with very good harnesses for people who work in the legal industry. What makes it powerful is context. Because it was built from our department's actual work, our policies, our playbooks, our institutional patterns. Claw doesn't just help with general legal work. For us, it helps with anthropics legal work.

13:29 And it can do the same for you. Again, I'm not a technical person. I've taken one computer science class in my life. The textbook was called C++ for UPS++. This was in the 90s. but my entire coding education, again, two sticky notes. I pushed my first GitHub pull request. Contribute to this project. If a lawyer can contribute code, your legal team can build on top of this. So what's exactly what's exactly changed here? The the goal was always to turn legal from the department that slows things down into crossf functional thought partners. When claude absorbs the information dense throughput work, the monitoring, the synthesis, the first drafts, then we're spending our time on judgment instead. complex negotiations, novel legal questions, the work that actually requires a lawyer. The throughput got faster, the work got better, and we got to keep our reputation as a legal department that can keep up with the pace of innovation in the AI era.

14:34 I want to welcome Alain back up to do another demo, walk us through a legal use case here. >> Thanks, Mark. >> Yeah. >> So, I think the the use case that you described is one that's a really great one to show. So again, thinking about kind of that cold start, where do you get started here? And the Jira use case is just so broadly applicable and something that I think a lot of teams can really take to heart. So in co-work, we're going to start with a simple analyze skill, right? And when we choose this analyze skill, Claude is actually going to prompt us with some information. Do we want to analyze customer turn trends and patterns or sales pipeline? These are some of the more common skills for analysis, right?

15:10 In this case, we obviously want to do something else. We want to work through our Jira backlog and see if we can kind of define any insights from it. So going from the beginning, we're looking to analyze the unassigned tickets in our legal ticket space in Jira and create a visual output. this is specifically going to help us with recommendations for best next steps as well. We don't want claude just to basically do simple data analysis. We actually want to use that intelligence of the model to say what would you do next? What is kind of the next best step here? And we're going to go ahead and run this. This analysis is going to connect to again as many MCP servers as we need. In this case, specifically the Jira MCP server. And it's going to take in all of the specifically unassigned tickets in that legal ticket space. Run that analysis and then create an interactive dashboard that's going to help us really understand what is going on here. As it's going through, it's going to build out its to-do list. So you can see the progress here. getting to the Jira cloud ID, fetching the unassigned tickets, analyzing and visualizing, and then generate that recommendations and final output. And through this process, you'll see that it's pulling in live those Jer tickets. So, general council review needed for web for updated website terms of service. I'm going to skip the analysis part of this and go straight to kind of the end result. What this looks like here is an example of what that dashboard could look like. So we're looking at the total unassigned tickets by category priority distribution tickets by requesting department priority and category heat mapping.

16:35 Mark, looking at this, is this kind of familiar? Does this look similar to what you'd be seeing? >> Yeah, absolutely. I I know this is dummy data, but it looks quite familiar to the exercise I did where we identified NDA tickets as as being top of the list. And >> what I really loved about it was that Claude suggested next steps for us how we could automate. So I think as you see there it's it's not just telling us data about where the backlogs might be but it's suggesting next steps and then I can have a conversation with Claude about how to attack that and build workflows to solve those bottlenecks we talked about earlier.

17:10 >> Cool. Love it. >> Excellent. Okay. now I'm going to bring up Matt Picella, one of our product managers. actually the person who taught me how to use GitHub. Matt's going to show you how stacking multiple plugins together has completely changed how he works. Matt, come on up. Thank you, Mark. There you go. Hello everyone. My name is Matt Picella and I lead our enterprise apps product team here at Enthropic. Today I'm going to tell you how I, as a product manager, use cloud co-work and plugins to build cla coowork and plugins.

17:49 So I'm a product manager. Product management is a very heavily cross functional job. On any given day, I'm doing a little bit of everything. Little bit of data analysis, a little bit of sales, little bit of engineering, a little bit of everything. My days are spent on 18 different things, mainly in meetings, which means that the real work happens in between the cracks of those meetings. I'm constantly getting assigned action items, assigning action items, following up on tasks, and fielding Slack pings and emails. It can be a little overwhelming. I'll be honest, the failure mode isn't that I can't do the work. It's that things just fall through the cracks and I I lose track of things.

18:32 Even if I had every task written down in a nice list, it's just every task is kind of loaded with context. A simple sentence like write the PRD for plug-in management means a lot to me. But in order to be able to do this, you need to know a lot. What are plugins in the first place? How do they fit in with my broader product? Who else needs to know I'm working on this? How does it fit into the broader company strategy?

18:58 Who are the engine leads I need to align with? And who are the lawyers like Mark that I need to talk to before it launches? Until recently, Claude just wasn't all that helpful for me as a product manager. But recently, plugins have really changed the game for me. And I'm going to show you a little bit about how it's done that. So, this is really a story of stacking four plugins on top of each other. And I'm going to walk through them one by one. First, and perhaps my favorite is the productivity plugin. The productivity plugin is my lifeline as a product manager, and it allows me to do two things. First, tasks, what I need to get done. In previous jobs, maybe this was scribbled down on an Apple note or a piece of paper on my desk. Now it's a structured list that Claude and I can maintain together and really divide and conquer.

19:47 Second is context and memory. All of the institutional knowledge that lives across the company that I need to know in order to do these tasks before that lived basically just in my head and now it's a structured file system that Claude can help me curate over time. Now, at the end of every day, I just sit with Claude and we run through everything together. What did I miss? What's still open? Are there people asking me for things that I urgently need to respond to? Claude acts as my personal chief of staff to make sure that I don't miss any tasks or leave any crossunctional stakeholder out on a limb.

20:26 Now, the second plugin, the data plugin, another big part of product management is informing your road map with real data. In a previous world, I might have written a bad SQL query, spent hours making a dashboard that only I ever looked at, and it kind of worked. But, you know, the answers I would get would often be wrong. They'd be directionally correct, and it would just take me way more time than I wanted to. Now, with the data plugin, Claude can help me with all of this. Claude can write the SQL queries way faster and more accurately than I might have otherwise. It can execute those queries without ever leaving co-work. It can help me visualize those results in easy to look at dashboards like you've seen from the other speakers. Plus, it can just answer questions like, you know, I talked to a data scientist or something like that. I can just answer questions that I have in real human language.

21:19 Third is the sales plugin. For any PM, staying close to the customer is essential. You need to know who your current customers are and what they're asking for. You need to know which customers in the pipeline, what they're asking for so that you can close deals. And my job is just significantly better when I have a clear window into what the sales team is seeing. The sales plugin gives me that. I can see who's in the pipeline. I can see real customer feedback and I can use all of that to inform and prioritize my road map and the features that we are building. Means that when I sit down with my sales counterparts like Eleanor, I'm not starting from zero. I already have context on what our customers need and who they are and their names and all the pipeline status. This means that we can focus the conversation on what matters most to maximize value for our customers.

22:09 Now, last but not least is the product plugin, the real bread and butter of what I do as a product manager. Vision and strategy setting, roadmap creation, writing PRDs. Previously, Claude just wasn't that great at doing this. I'll be honest, it was never Claude's capability. Claude's always been smart enough. What it was really missing was context. Think about what we've stacked up here. The productivity plugin gives both a list of tasks that are on my plate and all of the organizational context that helps Claude understand what's going on. The data plugin gives it real analytics, access to our data dashboards and warehouses, so it can, you know, query real things in real time. And the sales plugin gives it customer and pipeline information so that it can connect its PRD to a real customer set.

22:59 So now when I ask Cloud to write me a PRD, it's not just giving me some here are the five things that usually go in a PRD like it might have, you know, two or three years ago. It's actually helping me solve the hard problems that my work requires. It's a thought partner for me tightening and hardening my product strategy. It knows the product landscape of my company. It knows what customers are asking for. And this has really really changed the game on what it means to be a PM.

23:28 So when I say that Claude makes me a 10x better PM, I want to be really specific about what that means. It does not mean that I'm suddenly writing 10 times the number of PRDs. It's just that the things that used to slow me down or fall through the cracks just don't anymore. I stay organized. Things don't slip. Claude has all of the institutional knowledge, the data, the sales, the customer information, all of that to turn my ass into real outputs that really help customers. Plus, I want to say that every piece of this is getting better by the day. Claude is getting smarter. Our plugins are getting more powerful. Co-work is getting easier to use. And so, however easy it is to use today, every day it will get easier and better and better.

24:12 Now, with that, I'll call Alone up to help out show off some of this stuff with the demo. >> Thanks, Matt. So, if we think about like the productivity plug-in being in a lot of ways the heartbeat of the work that you're doing, I wanted to demonstrate what that could look like and and do it in a couple of ways. So, the first is we're going to look for our update skill. Our update skills specifically looks at the work that we've done today and is going to update some of those to-do lists, some of the things that you've been working on to ensure that it's got all of the context that you need. Things that are already complete, get to be completed. Things that are still to be done are going to stay on that to-do list, maybe be rep prioritized, right? And so when we choose this update skill, it's going to ask us again some information, right?

24:52 Do we want to do a quick sync of current tasks, a deep refresh, or this comprehensive update, which is the one I'm going to choose here. going through this comprehensive update, like I said, is going to connect to basically all the systems that you're working in day-to-day. It's going to connect to your Slack instance, your Gmail, your Google calendar, anything that you're working in on a regular basis to ensure that the system and cloud now have all of the information it needs to ensure that it's updating what you're doing. Now, while this is running, and this will take a few minutes, I'm going to go ahead and pull up what that dashboard could look like. So, this is an example productivity dashboard. It's broken into a few different sections.

25:27 Matt, why don't you walk through kind of the difference between tasks and memory here? >> Yeah, it's a good question. So, so like I said before, tasks are really just a way for both myself to stay organized and a way for Claude and I to have shared language about what's on my plate. So Claude can read these, I can read these, Claude can write them, I can write them. It's just a way for both of us to stay organized together. In the memory tab, that's really all of the organizational context. And this was completely created by Claude. It's a set of projects, people, terms, all sorts of stuff that Claude needs in order to help out on my tasks. So, when I say, "Hey, can you ping Stephanie to get that PRD on that project?" It knows what each of those words mean and can actually help me.

26:10 And so if we go back to the actual cloud instance here, you'll see it's connecting to all of those different systems and it's going to basically go in and as Matt said, this kind of read and write process is so important when it's syncing through these systems, when it's running through to make sure that it's got all the information that it needs. It can then write to that productivity dashboard and make sure that it's staying as uptodate as possible, ensuring that Mac can do his job as efficiently as possible.

26:34 Cool. And with that, I will turn it over to Eleanor who can talk through how Claude Co is helping her team on sales. Thank you, Matt. It's my turn. This sales team is an incredibly heavy user of co-work. And I'm going to use this session to walk you through how our AEES and our managers, our account managers, applied AI, BDRs, we all use co-work every day. So think about how sales actually works today. Your CRM holds the deals. Your Slack and your teams holds the institutional knowledge.

27:20 Your inbox holds the relationship. Your cell phone holds the relationship. Call recordings hold the insight. And for better or worse, spreadsheets still hold the forecasts. And every day, your reps move in between these systems every morning, every afternoon, most evenings. They're copying, they're pasting, they're updating, they're inputting data and operating software instead of executing deals and running sales. The average rep, at least in my experience, spends more time documenting the work than actually doing the work.

27:53 So, what if we flipped it? What if sales didn't have to go to their systems and their tools? And what if the tools and the systems and the context came to them? That is what the sales context, the sales plug-in unlocks. You give Claude a sales skin. You connect your HubSpot, your Salesforce, your Dawn, your Gmail, your Slack, your Teams, and suddenly Claude isn't just a chatbot answering questions. It's working alongside you as you run your day after day after day. have a call. Claude lodgs the activity, updates the opportunity, drafts the followup, and sends it.

28:32 Prepping for a meeting, Claude pulls the account history, the recent interactions with the customer, the product usage trends, the attendee backgrounds. One brief, ready when you are have a pipeline review, which we know you all have every week. Cloud will audit your deals, your Salesforce, and your HubSpot. It'll flag what's gone stale. It will prioritize what needs attention. It'll surface missing fields. And if you teach Claude once, it runs the workflow alongside of you every time.

29:02 Let me get more specific. What we've specifically done is build five stills that encode how we've seen elite sellers work. The morning briefing, the call prep, customer follow-up, competitive intelligence, and create an asset. As a seller, you're pulled in a million different directions every day, internally and externally. What your morning briefing does is ensure you start your day off productively with the right context and the right orientation. It pulls the day meetings with context.

29:34 It flags your open and active deals that need attention. It tells you which emails to respond to first and gives you three prioritized tasks. It takes 2 minutes and your entire day is organized. The second still is call prep, which is core to the AE workload. You have meetings all day, every week. And Claude can assess who you're meeting, what they care about. It can look at all of your previous interactions. It can look at conversations on that deal channel and Slack, and it can recommend what questions to ask. Before a customer meeting, your best reps would spend 30 minutes combing through LinkedIn, Salesforce, Gmail, their co-workers.

30:16 Now, it's instant and it can happen before every meeting. Now, after the meeting is the follow-up, which as a leader you spend a lot of time worrying about, has this fallen through the cracks? That is what the customer follow-up still is designed to address. After a call, you paste your notes or your transcript into Claude. Claude will extract the action items, draft the customer follow-up, and put together and synthesize the internal summary. If your CRM is connected, which we strongly recommend, it offers to log everything and create the tasks. This isn't just a better rep experience. This is a better customer experience from prep through the call all the way through the follow-up.

30:57 The fourth is important and it's something we all think about a lot. Competitive intelligence and battle cards. We've created a competitive intelligence skill that's part of the sales plugin that generates an interactive battle card that includes a comparison matrix, talk tracks, and landmine questions to plant and ensures that this is available consistently and updated for your entire org, especially as the competitive landscape evolves. This cannot just be a static document that's maintained somewhere in a CRM.

31:26 This is a living, breathing document, and you want to ensure your whole team is leveraging it. This last still is my favorite, create an asset, and it's a bit of a wild card. This was inspired by someone on our team and has proven incredibly valuable across the board. What you do here is you describe to Claude, who's your prospect, what are their priorities, what do they care about, and what do you need at that moment in that deal for that account. Claude will put together a polished branded landing page, an interactive demo, a customized onepager, or a beautiful deck. This is custom sales collateral that used to be prioritized only for the most strategic deals, usually only after a few installations and required a sales manager, a PMM, and a designer. Now, it's available to any c for any customer for any deal, and any rep at any stage in the sales process. And while you never quite know what you're going to get, you can workshop it with Claude to ensure that it'll be valuable and move the deal forward.

32:27 With the sales plugin, you remove one of the single biggest barriers to AI adoption, which is the cold start problem. A rep can paste in notes and did a structured summary. You can upload a CSV and did a weighted forecast. You can describe your pipeline and did a prioritized action plan. Zero setup required once you have this still set up, the plug-in set up. But the real unlock comes when you connect your systems. When you wire up your CRM, connect your email, connect your calendar, connect your enrichment tools, and Claude moves from working with your data and answering questions to working inside your workflows on your behalf. A co-worker embedded alongside of you to support your customers every single day.

33:11 You can get started in minutes and go deeper as you connect more and more of your stack. This isn't about replacing reps. It's about returning them to the work that they were hired to do and humanizing their day-to-day. We have a brilliant and dedicated sales team at Enthropic. But there is a heavy operational burden. And when Claude handles that operational burden, the data entry, the context gathering, the systems updates, the value of the work goes up significantly that they're able to deliver. And it's not just the rep experience. I really want to reiterate this. It dramatically improves the customer experience as well. Reps get hours back to work the deals that really matter, to prepare and execute on a complex negotiation, and to deliver a more personalized and attentive customer experience.

34:01 They can better manage all of this. And selfishly, as a manager, you spend far less time chasing reps for CRM updates. My team, since we've implemented this, spends so much more time coaching, listening to calls, working with their reps, riding shotgun with them in deals. I'm personally spend far more time on the strategic questions of our business and hiring and recruiting and all of the problems that only a human can solve. There's a compounding effect, which is what makes this so powerful. Every still you build, every workflow you teach, Claude, compounds across your entire organization. We're salespeople and we like to talk and we like to learn from each other. And this enables us to find every best practice and ensure that it's harnessed by the entire team. The best call prep becomes everyone's call prep.

34:52 The best discovery framework becomes what everyone does in their onboarding. Your top performers follow-ups becomes everyone follow-ups. Pipeline review methodology is no longer tribal knowledge. It's available and used by everyone. For years, sales team have shaped their processes, their days, their habits around their tools. They've learned their quirks. They've adapted to them and they've operated within their restraints. What's exciting is that changes today and the approach becomes about what the salesperson wants to bring to the deal because the best practices, the operations, all of that is out of the box with the sales plugin with Claude. Your systems come to you, your tools come to you, your deal context comes to you, the tribal knowledge comes to you. You no longer have to grab time with that first rep on your first week because it's all just part of how you work.

35:48 The workflows run alongside you and sales again has found its coworker and it enables your team and yourself to reclaim the work that really matters. Now, I'm a saleserson, so naturally I'm raing a lawn back up to do a demo of one of these live. >> Thanks, Eleanor. And of course, I'm going to make my salesperson happy and demo her favorite thing here. So, we're going to look at the create an asset skill. if we set the stage quickly, we just had a meeting with one of our current customers. we are positioning an upsell with them, but we also discovered there's some competitive pressure here. And we heard during that meeting that one of the things that our champion needs is an asset or some information to be able to bring to their executive leadership team to be able to of course communicate, you know, how well our relationship has has developed and our partnerships has deepened as well as what is the ROI framework and value creation that we can continue to create for our customer over the course of our partnership as we continue to deepen it even further. And so with our create an asset skill, I can go in here and paste in some information, about what I need. And again, this is really simple. I simply want to review the next steps for the customer for Pinnacle Investment Partners, and pull that from HubSpot, make sure that we're creating that interactive page that I can send to the customer. I'm not putting in a ton of detail here. I'm not putting in a ton of information about what I need because our systems are connected. my AE is amazing and put in really detailed next steps into our CRM.

37:17 And so I don't need to reiterate that to Claude. I can go in and just say, look at those next steps, pull what we need, and then be able to build that asset. Claude's going to go in, connect to HubSpot, connect to our CRM. It's going to be able to go and start to build that asset out. And what we get at the end of the day is that customer ready asset. In this case, I'm going to flip right to it. We're looking at a web page that is interactive and that can be delivered directly to the customer and shared around as needed. So we talk about our expanding a proven partnership to enterprise scale. We've got our current investment from the customer to us as well as the partnership duration. We walk through a timeline of you know where we started and how we've gotten to this point. We can look at the value delivered right. So at this current kind of ARR what's the value that we have really created for our customer 40% faster report generation 3x portfol port portfolio visibility 100% compliance coverage these are meaningful and tangible results right we have our operational impact and our industry benchmarks to ensure that we understand how we're basing this off of the current state off of those benchmarks and then crucially here we can go to that ROI framework what's really cool about this ROI framework is this is totally interactive right So we have some assumptions in here. Number of advisers at 40, average hours saved per adviser per week, etc. And this gives us kind of that annual productivity savings, the net annual ROI, and kind of an eventual payback period. But what if the customer wants to play around with this? What if they say, well, we're actually thinking about growing our team to about 60 people. And we have, you know, we're a little bit more skeptical about the benefit that you guys provide. We don't think that your assumption is correct.

38:56 We're going to drop this down to maybe saving six hours per week. But we do think that fully fully loaded hourly cost is going to jump up to call it 200, right? We can actually see in real time how these results and how this value creation is going to change based on whether the assumptions that we made are in line with what the customer expects. Elanor, what was your reaction the first time an AE brought you something like this? Like how do you look at this and say this is changing how sales works? We had an AE demo this actually at our sales takeoff and my immediate reaction was we need to build a still because everyone needs to be using this which is what's so powerful about these plugins and with stills. As soon as you identify a best practice you can encode it into everyone's best practices.

39:37 >> Awesome. >> That's a wrap for this session. I hope you all enjoyed it. Thank you so much to Alain for all of the incredible demos, to Mark, to Lindsay, to Matt for showing us what a non-coding lawyer can produce, and a PM who stacks plugins on plugins on plugins can achieve in a day. These are not edge cases. These are things that all of our teams do every day. This is just mundane daily work simply done better with Claude. If any of these demos or sessions sparked something for you, please reach out to your sales team. And thank you so much for your time today.