Name: Inside Claude for Financial Advisors
Uploaded: 2026-09-18T16:32:33.359Z
Duration: 54 min 28 s
Description: Inside Claude for Financial Advisors

Transcript for "Inside Claude for Financial Advisors": Good morning, and welcome to our webinar this morning. Thank you so much for joining. We're gonna start with some brief introductions, and then we'll go into our agenda this morning. So first off, an introduction to myself. I'm Belinda Neal, the managing director of financial services here at Anthropic. Passing to you, Peter. Hey, everybody. This is Peter Nolan, head of asset and wealth management at Anthropic. So are you there? Yeah. Sorry. Just having some audio issues. Hey, everybody. My name is Drew Parker. I head up the RIA segment at Anthropic, but really excited for this webinar today. Excellent. So welcome, everybody. First and foremost, thank you for joining today. We've had a really exciting launch week of our new Florida financial advisers product that we delivered on Monday. We've had some really great headlines and some really great activation with our partners that were part of the launch. So first and foremost, thank you to our partners, and hopefully you've seen some of the press coverage and you're excited to learn more. Our goal for today is to really go through this product in a little bit more detail and have time for question and answers. So we're really hoping to show you in detail what this product is, what it can do, and get your questions answered. First off is a quick start. Since we launched on Monday, over 30,000 people have installed the financial for financial advisers plug in. So we're seeing some really struck strong engagement and adoption so far, and we really look forward to your feedback. Just in terms of the webinar today, a few housekeeping things. Firstly, a recording of the session will be distributed via email in twenty four hours so you can watch it again as you wish. Secondly, we will have a q and a section so questions can be submitted at any time in the q and a tab in this webinar portal. Please think of them in advance and send them as the webinar goes along, and we're gonna spend some time going through curated questions. Please also feel free to offer any questions that are of particular interest to you so we get to the ones that are most useful for this group. And then lastly, we will have a feedback survey after. We would love you to fill that out and share some feedback with with with us, and feel free to send any questions our direction as well. So we're gonna start off with a quick poll. Firstly, we'd like to ask this whole group, how many hours a week go to meeting prep and follow-up for a financial adviser? So Taylor's gonna queue a poll, and we'd love you all to answer this question. Just take a minute. Okay. Please react with your feedback. Okay. Taylor, I don't know if you can share more of the results or we just see under five hours. Great. So it looks like overall, a number of you have voted 367 people say under five hours, seven seven four say five to ten hours, three seven three eight five say ten to twenty hours, and 98 say over twenty hours. So there's a fairly large distribution here of how people are experiencing prep time that goes into the adviser workflow. So that's a really interesting data point that we will start with, as we go through the webinar. Before we go into the details of this, we wanted to share this launch offer with you in case you weren't here at the end. Right now, we are offering a $4,800 credit for those of you who subscribe to the enterprise plan. The advisory sorry. The financial advisers firms are recommended to join the Claude enterprise plan to ensure we can deliver to you the enterprise features needed for Claude for your advisory practice. So we encourage you to use the sign up form to get the best version of Claude for this product. And if you, register by the September 30, you will receive this credit, which is good for six months. Okay. Let's go through our agenda today. So first off, Peter Nolan, who leads our Ascendant Wealth Management business, will go through the opportunity for advisers. Next, Drew Parker will go through a live demo of the product itself. Then we're gonna go through some live q and a with our speakers today, and then we're gonna go through next steps. So without anything further, Peter, I'll pass to you. Awesome. Thank you so much, Belinda, and thank you everybody for tuning in. Many of you may have been on the ground at Future Proof when we announced the product this past week, but I'm certainly excited to share this with the rest of you. Taylor, if we could flip the slide. I'll start with why we're here and what problem we hope to solve. Infropic's mission is responsible AI. We wanna make sure that powerful AI is used in ways that help society writ large. In my view and on my team, a major problem is financial wellness. And I think if you can solve financial wellness, a lot of the downstream problems from it can also be solved. So think affordable health care, affordable housing, affordable education. There's a common denominator here, that can be solved if we can help people figure out finances. I think a primary cause of a lot of financial wellness issues is something we call the advice gap. So just sixty one percent of workers expect to be able to retire comfortably. Sixty three percent of adults could cover a $400 emergency in cash, and that's likely because only four in 10 Americans have one of you a financial adviser. So we see this as a supply problem. Taylor, if you could flip the slide. And so the question then becomes, as an AI company, how best to solve a supply problem? We wanna get as much financial advice out there as we possibly can because we know it solves investment outcomes. Flip the slide, please, Taylor. A natural path might be to create a direct to consumer financial advisor to offer advice through Claude. That's not the path we're taking. And the primary reason for that is we're students of history. In my estimation and the rest of the team at Anthropic, we've seen financial innovation come to market many times. And each time, it is espoused to threaten financial advisers but eventually becomes an arrow in their quiver or part of their arsenal in helping their clients. So let's let's take a quick, pass through history. In the nineteen seventies, the index mutual fund was launched as a low cost basket of stocks. Why do you need a financial adviser when you can get access to the markets in such an efficient way? And so there was a big, you know, apparent threat. But today, 66% of ETF assets sit in adviser managed accounts, and financial advisers are the primary purveyors of this technology that ultimately helps investors reach retirement outcomes. Fast forward to the nineteen eighties, May Day and discount brokerage arose, so did online trading. All of a sudden, consumers could log on directly, could access research for free, and could execute on their own. So, again, why do you need a financial adviser? Well, it turns out the primary users of these platforms or power users were financial advisers themselves. Schwab noticed that in the late nineteen eighties and pioneered the RIA custody model, serving those power users who were serving other people. And it just happened that they birthed an industry that today is significant. Schwab safeguards about five point five to six trillion in your clients' assets and is a major supporter of the financial adviser ecosystem. The final trend in the 2,000 and tens robo advisors rose. Everybody remembers this. Financial advisers were put on notice. It just so happens that today, the largest robo advisors by far sit behind you. They're there to automate portfolio management so that you can use your skills for higher level tasks like financial planning, relationship management, behavioral coaching, etcetera. Flipping the slide, please. So the theme here is every wave ran through advisers, and we think this one will too. That's why we're building for you. So this isn't called the financial adviser. This is Claude for financial advisers. Flip the slide, please. The problem we wanna solve is this. Today, just 20% of your time is spent with clients. The rest of it is spent on things that we think we can automate using AI. Flip the slide. Because of the inefficiencies in today's workflows, most of the work go to a subset of families. Again, going back to a supply a supply problem. Flip the slide. And so we see the system is broken because the financial adviser needs to be the integration layer between all the different components of the tech stack. Independent advice is a very important space. A lot of advisers go independent so they can get choice. Choice means that they have many things to select from from a tech stack perspective, but the problem becomes integration. And that's what we think we can solve. Flip the slide. Claude for financial advisers will be the orchestration layer for your tech stack moving forward. What this does is Claude will sit between you and the various components of the tech stack to make sure it all sings. John Beatty, head of finance or head of adviser services at Schwab earlier this week, said Claude would be the chief orchestration officer of the RIA industry moving forward. That's exactly the vision we hope to achieve by partnering with you. And, flip the slide, please, Taylor, the leading providers of wealth management technology in the industry today. We're very proud to partner with BlackRock, Charles Schwab, iCapital, Vanguard, all the leading portfolio accounting systems, market data, CRM, and the core software stack. Claude has MCP integrations into each of these, and Drew Parker is gonna show you in the coming minutes how you can stand all of this up, integrate it into Claude so that Claude can orchestrate across these components and automate workflows end to end. Drew Parker, come join me. How's the audio doing now? Still good? It's good. Amazing. Hey, everybody. Okay. Taylor, if we could flip the slide, please. Okay. So what actually comes included in this plug in? Well, it's, two things. One is all of these MCP connections that Peter was just talking about. What does an MCP connection mean? It's just an integration to the tool that you're actually using. So we have integrations to, financial planning software, portfolio management software, CRMs, the, like, core stack that you use as well for just software, like your email, Google Drive or SharePoint or anything else that you're using there. And on top of all of those MCP connections, we prebuilt skills that we are shipping, in the plug in as well. For those that aren't familiar with a skill, a skill is something that you can run-in Claude Cowork where you can make Claude's output deterministic. So you teach Claude the exact order of operations that you want it to operate within, and then you give it, here's the actual deliverable that I want you to send me. We'll go through each of these skills in-depth today. And then lastly, on top of that, we are obviously very capacity constrained at Anthropic. Our most scarce resources are people. And so we have prebuilt an onboarding workflow within the plug in when it's installed, and I will show you all how to invoke that as well today. But, Taylor, let's go to the next slide. What are some of these skills? We try to bucket everything out into just, like, different aspects of the adviser's day. So first, we have pre meeting. You've got a bunch of different systems that historically you'd need to swivel chair between. How can you point Claude at each of them and say, Claude, here's the, process that I want you to operate within and the actual deliverable that I want you to send me. We'll go into more detail into all of these, so I don't wanna spend too much time here now. But this is the overview of the skills that we'll be presenting. Next slide. On top of that, how can you trust Anthropic with all of this data being plugged into Claude across your most sensitive client data from Social Security numbers and tax returns and performance data and all the disparate systems that you're using? This is because and why Belinda Neal was mentioning earlier that this should be used on the enterprise license of Claude because Anthropic is contractually, precluded from training on any of your data within the enterprise license out of the box. There's no separate agreement you need to have to enable this. There's no addendum to a contract that you need. Anybody on an enterprise license, is perfectly isolated within their four walls of Claude. What does that mean? All data that you directly prompt Claude with and type into a chat bar, all data that is brought into the session via Cowork and the integrations that you set up across your systems, as well as Claude's outputs back to you. So everything that Claude returns to you in your session, that data is yours and yours alone. Nobody at Anthropic can see it. We can't train on it. We can't package it up and sell it. That is yours and yours, to keep. On top of that, we have spent a ton of time making each and every one of our products enterprise ready and giving the level of admin controls and, other features that you need to make sure that the permissions are in place so that Claude acts in a, compliant manner. So first off, there's human in the loop approval on any action that Claude is going to take. So before Claude actually uses its hands and goes and does anything out in the real world across the systems that you've connected, Claude is going to tell you exactly what its plan is, then the human can actually hit approve, deny, or edit the actual course of action. That is out of the box, the the, the setup that we have for everybody within their Cowork session. On top of that, and we kinda already hit on this a little bit, but the on the data security front, all of our Claude for work plans, so this includes the team and enterprise plans. We don't train on any of the data again. I think the main reason we are prescribing everybody to use the enterprise license over this connector is those RAs that have gotten comfortable with using Claude on a team plan from a compliance perspective have been able to do so because they are scrubbing PII before it even hits Claude, and they are they haven't connected it across the broader stack. And so having the need for an audit log to point at your archiving or ediscovery software to keep record of all the things that Claude is doing for you wasn't really necessary. Now with this plug in where Claude can affect changes on your CRM based off of updates that it pulls from your portfolio management software, your tax and estate planning tool, or anything else, the these are material changes that should and need to be tracked. And the only way to do so is on the enterprise license. The for context, the team plan does not have our compliance API to point at another system. Next slide, please. Okay. So with all that, let me get my screen up here and walk you through the tool. Can everybody see this okay? Great. Okay. So first off, where do you actually go to install this? Everybody within their Claude account can install the plug in today. If you toggle over to Cowork within the desktop or claude.ai, site, and then come down to the left bottom left of the screen and click on your, profile icon, then go into the customize section and hit plugins. You'll see a financial advisors plugin. You can go ahead and click on that and hit install. That's it. It's as simple as that. There are no other configurations or steps that you need to do. Once this plugin is installed, it will be saved on your Cowork, and you can run the onboarding workflow. There's gonna be a bunch of different buttons and toggles on this screen. We frankly found it a bit confusing, and so we have packaged everything up in this onboarding workflow that I'll run you through right now so you don't have to worry about anything else on the plug in installation screen. Once the plug in is hooked on, type in backslash onboarding, and Claude will onboard you to the tool. It will first look to gather some context on your role, the the type of firm that you're working at, and any other details that you provide it. The more information you give Claude, the better. The more context that it has, the more applicable it can be to your role in your firm. So it'll ask you what's your role, what kind of firm are you working at, etcetera. Next up, it'll ask you what are some of your day to day processes or workflows that are causing you, headaches. You can go and select the ones that are relevant. And then lastly, it will ask you to tell Claude what your tech stack looks like. So you can go through and select all the tools, CRM, portfolio management software, alternatives, portfolio analytics, select all that are applicable, financial planning tools, tax and estate, etcetera. Select all that apply. Then Claude will go ahead and surface the actual connectors to stand up all of these integrations. You will press connect on each one. That will take you out into a separate window where you'll in to each of the given tools that you're connecting. Then Claude will go ahead and give a bit of context on what the plug in contains. I'd recommend asking Claude, hey. Give me a brief overview on each of the skills. If you wanna go further in-depth or more in-depth, I should say, ask Claude Claude from start to finish. Tell me exactly in plain English what the pre meeting skill does or what post meeting does and go through each one. But when you're ready to actually run your first skill, you can do so by either doing the backslash command or just doing it in plain English. So let's say, for example, we've installed the plug in now. We're doing some work over the weekend to prepare for a Monday morning meeting that we have with Jane Jetson. What do we do? Well, in Cowork, come over to the chat bar, and you can either do the backslash pre meeting client name or in just plain English. You can tell Claude what you wanted to do. When you send that prompt, the first thing Claude will do is take the client name and find it in your CRM. If there are any potential duplicates or it's not a 100% sure who the client is, it will ask you to confirm. Once Claude has pulled the household or client, it will then pivot to all the different systems that you're looking for. So first, it'll just start in the CRM, identify any notes, that you have on the account, any action items that you need to discuss this time, follow-up items from last time, etcetera. Then it'll go to your email. Go through all co correspondence you've had since the last time that you spoke to the client. Then it'll go to your portfolio management software and pull the performance data. Claude can actually pull, position data from each, household or account that it has. Then it will go to your financial planning tool, get the financial plan, and all the other systems that you've connected to ultimately generate a meeting brief for you that will be displayed as an artifact right here on the right of the screen. That is the ultimate brief, though, that Claude will generate, and we'll get into at the end how you can actually customize the the deliverable. Next up, let's say we just wrapped the meeting with Jane, went well. We wanna log all the updates to CRM. You'll need to transcribe the meeting. If you're using Zox, there's a connector built there, so you don't need to do anything but have the connector stood up. If you're using another notetaker, just go ahead and drag and drop in the transcript, to the the chat that you open up. Claude will read through the transcript and identify any next steps, action items, or things that need to be discussed further, and then package them all up to you as updates listed as an artifact on the right of the screen. Claude will not make any of these changes or updates in your CRM before you approve. So once you read through everything that Claude has surfaced and hit approve, Claude will then go to your CRM and make all those changes accordingly. Now let's say we we don't quite remember what Jane was saying about Fred Flintstone and the potential referral there. You can just ask Claude. Claude has a transcript. It will return back to you what Jane offered. And then you can ask Claude to schedule a nudge if the intro hasn't happened after x amount of time, but just a a little use case of how you can, like, iteratively work with Claude and basically have this become your adviser cockpit. We couldn't share a private outside. Love it. Done. Next up, prospect intake. So let's say fast forward three weeks, we've now had an intro call and discovery call with Fred Flintstone. We've asked him to send over, all of his financial documents and everything else that we're asking for. Fred has finally sent it over in an email. We can ask Cowork to go and pull all those documents directly from the email that Fred has sent, or we can paste them in and then run prospect intake on him. What this skill does is it looks to take all of the different file formats that I can only imagine that clients send over for all their financial statements, whether that be PDFs, screenshots of, iOS app dashboards, and however else things are sent over. But Claude will parse through everything, standardize the format, and produce a couple of things for you. The first is just a prospect summary doc that will take all of his accounts, positions, holdings, all the data that you've had that you have over that was sent over into one consolidated place and then deliver two deliverables to you. The first is a what to expect doc that you can send to the client, basically detailing, here's what you sent me. Here's what we can do going forward. And the next is if you're working with an analyst team to actually develop the investment proposals, that are gonna be created, you can basically hand over this this doc directly to that team to have them run the proposal. On top of that, there is a compliance skill that comes in all this. We'll demo it in a second here, but, basically, anything that's client facing, Claude will run past the compliance check to ensure that all the the rules and regulations that you need to adhere to are actually being adhered to. Next up, estate and tax brief. So this runs off our wealth.com, connector. The use case, and there are many, but to highlight one is, let's say, we've got the estate and tax doc listed out in wealth.com. We wanna just identify any discrepancies over what the actual plan says versus what is reflected in all your various systems that you're using, like portfolio management tools, financial plans, etcetera. You can tell Claude to run the state and tax free skill. If you see up here, for example, we didn't specify the client. So Claude doesn't know who to pull it from. It'll ask you to specify. Once you give the client, Anthropic will check wealth.com for the actual estate and tax plan, surface it for itself, and then go across every single one of your systems to identify if there are any discrepancies. Claude will return back what looks good and other things that need potential discussion over a meeting or a quick call and generate just a full brief doc for you that you can send to either the client or you can just use yourself to keep track of anything that needs attention. Claude will then offer to get all of this reflected in your CRM as well. So you can keep the the meeting notes or action items that need to be discussed during, like, an upcoming quarterly review meeting or just a quick phone call to the client. Again, with any updates to CRM, those need to be approved before those are actually reflected. Cool. Next up, portfolio rebalance and alts skill. So the use case here or one I'd like to highlight is, let's say we get an email in from a a client, as I can imagine happens often, asking, how am I doing versus the market? How am I performing? When you ask that same question to Anthropic, Anthropic will invoke the portfolio rebalance skill, point to your portfolio management software. If you're using multiple, Anthropic will toggle between both to identify which one the client is actually located in. Again, if there are any potential dupes or it's potentially confused on anything, it will ask you to clarify. Claude will then go and pull all of the position data for each of the accounts that you have connected for that household or client and then return the metrics to you. In this case, Morticia is up 6.2% on the year versus 11.7 on the benchmark. So what's causing the lag? Claude will look into it, and it's because Morticia is almost 21% in cash. That's sort of like the level one answer, but Claude will continue digging. And as you ask Claude to build the actual memo to the client as it keeps on digging, it'll say, you might wanna look at if any of this cache is committed or not. And this sort of pairs nicely with the the alt skill. This runs off of iCapital. Claude will then go and pull all the private book data directly from iCapital to determine how much of the cash is actually free. And there you go. So of the the 2,400,000 in cash, looks like 1.4 plus was an unfunded commitments, so it's not free. And there's your answer to the client. And so Claude will then generate the full brief for you that you can use to email back to the client indicating, yes, we're down comparatively in comparison to the benchmark. But here's why. Next up, deep dive on compliance. So I think the the best use case for this is, anytime you're gonna be sending a market commentary piece to all of your to your book of business or just any email, frankly, to clients, you can drag and drop in as a file the messaging that you look to be you wanna be putting out and run it against the compliance skill. This takes into account all the SEC marketing rules and regulations as well as other, regulatory rules. I would recommend that everybody also layer on their firm specific sort of guidelines. Each and every one of you will have your own slightly different interpretation of the rules, and I think that's great. But you'll wanna lay layer those into the skill as well. Once you upload the actual document, Claude will take a first pass at it, identify any things that potentially need review, and then recommend changes to you. Any points that need clarifying, Claude will always ask. It is specifically trained to not assume. And, ultimately, it'll deliver the check. And so I think how I envision a lot of firms using this is before even sending a piece to a CCO, running it past the compliance skill to identify all the potential red flags and things that can be cleaned up quickly to just decrease the approval time once sent to compliance. And that's it. So I think you'll notice each of these skills are fairly generic in nature, and and this is intentional. We design these to be applicable to as many different people as possible. What we would recommend and would love for you all to do is make these skills your own. So how do you actually customize skills in Claude? Let's say, for example, on the pre meeting skill. It works well, but we don't wanna have to invoke the skill every time that we have an upcoming client meeting. Instead, we want Claude to, on a scheduled basis, just scrape our calendar, identify any external meetings, take the email, on the meeting associated with the invite, then take that, email address, paste it to CRM, identify the household or the client. And we want a meeting brief doc for every single one of our meetings the next day delivered to our email address or delivered wherever you'd like it to go right at 8AM. You can tell those updates to Claude, and Claude will work through with you how to actually go and customize or edit that skill. Claude will surface all the updates that it looks to make to the skill and then ask you to approve. Once you hit set it up, Claude will work through the updates, update the actual skill file, which is an MD file, to run off of every every subsequent time. So now every time you were to run the pre meeting skill, your updates would be included. So you no longer have to run it every day. Claude will just autorun it on a scheduled basis. But that is the demo. Let me stop sharing my screen here. Okay. So in terms of getting set up, like we kinda talked about, the skill is meant to be or sorry. The plug in is meant to be as self serve as possible to just enable firms to get the most out of Claude with without being dependent on a bunch of help from either partners or anthropic internally. Like I also mentioned, we are very capacity constrained, and so we won't have a lot of the same, like, white glove, hand holding throughout the onboarding and implementation process, and we recognize that as a gap. And so instead of just saying, hey. Figure it out. There's nothing we can do. We have spent a ton of time and resources enabling a laundry list of fantastic partners that are RIA specific. They have all the context. They know your workflows. They know your business, and they would love nothing more than to be boots on the ground help for enabling the team, implementing the tool, helping you customize each of the skills that we're shipping, even build your own skills on top of that, and refine and identify all the use cases that you'd like to work with, Claude on. For firms that are looking to go all in and do a wall to wall deployment and are willing to make a significant spend commitment to Claude on an annual basis, we do have some internal resources that we can give as well. We've got a state of the or a phenomenal applied AI team that can be looped in and spin up managed agents that you can run, and really go that next level. But, again, that is just sort of spend gated given the capacity constraints. And just to name some of the partners that we work with, Artiom, they were instrumental in actually building out this plug in with us. They developed all the skills and tested them off of each of the connectors that we have stood up. They're fantastic. Improve. We were at Future Proof with them. They've got a fantastic team. F two strategy, they are phenomenal as well, same as Fulcrum five. PWC, if you if you want a bit more of a a larger firm to work with, and their account is more. But if you're interested in working with a partner, feel free to reach out directly to them. They all have the context on this plug in and are ready to go. And if you'd like us to make intros as well, we can facilitate that too. But that's it. Excellent. Thank you so much, Drew. That was a really great demo, and, hopefully, everyone here enjoyed that. Next up, we're gonna stop going through q and a. We've received quite a lot of questions as you can imagine, so we very much appreciate your engagement. I think thematically I'm going to go through them, section by section and we'll try and cover the bulk of what people are asking. And if we don't answer your question or if you still have additional feedback, please use the feedback form at the end of this session to get to us so we can help create an FAQ that will be really useful, for this group. So I think the biggest question that I've received and seen through these q and a polls, Drew, is around the enterprise plan at 20 seats. So perhaps you can spend a few minutes talking about why there's a 20 seat minimum, why we need people to use the enterprise plan for this, what features does that unlock for a firm. If you only have four or five people, why a 20 license, is what we're what we're what we're, recommending people do, and then how the credit kind of overlays into that, offer. So why don't you take that first, Drew? 100%. So the enterprise license comes with a hard floor of 20 seats. Frankly, it's a bit of an arbitrary number, but there is no lever that anybody at Anthropic has to lowering that minimum seat count. The average RA does not have 20 people at it. And so I think when we were thinking about this launch, Peter, Belinda, and I spent a ton of time coming up with, okay. Well, we're gonna need RAs to be on the enterprise license to really use this tool or plug in compliantly. What can we do to basically get our AAs with one or just fewer than 20 people at it onto an enterprise plan without penalizing them from, like, a cost perspective for being a small firm? And so that's where we kinda put our heads together and came up with the launch offer. So enterprise seats cost $20 per month per user. It's paid up front for the year. That equals $4,800. And so that's where the the $4,800 usage credit really came from. It was, sure, you'll still need to purchase seats for the year $4,800, but then we will give you $4,800 in credits to offset the seat fee, entirely, on a go forward basis. And to give some more context as to why, this plug in should really be run on the enterprise plan is the team plan is designed for small and unregulated firms. It does not have access to any audit logs. It does not have RBAC controls layered in. There are frankly very limited admin and oversight capabilities in the plan as a whole. The enterprise plan is the opposite of that. It is designed for large and highly regulated firms. It has a compliance API that you can direct towards your archiving or seam software or your discovery. And on top of that, it has a full suite of RBAC controls. So you can actually set a model menu available to all of the people at your firm. You can also set model defaults. You can set spend caps, which is your lever to controlling cost. And I think it would also be helpful to just talk about enterprise pricing and how that works. Maybe we can do that in a in another question, Belinda. But, Let's just tackle it now. Why don't you go through what the estimated token cost would be and how that dovetails to the license fee? 100%. So enterprise seats, as I said, $20 per user per month billed upfront for the year. At a minimum, you need to purchase the 20 to get started on the enterprise plan. If you need to purchase additional seats at any time, you can do so in the admin, portal. All seats purchased in the future are prorated based off the amount of time left on the plan. So if you are six months into your enterprise license and you buy an additional seat, that seat would cost $20 times six months, times one for the one user. Usage is built separately at API rates. Our API rates are available publicly online. Just search anthropic API rates. It'll be the first link that comes up. To give some context over what usage fees will cost, I would say our average Coworkers users are gonna be between about 70 to a $120 on average using the plug in. Obviously, usage is incredibly variable. It just is entirely dependent on what you do with Claude. Your lever to controlling spend is spend caps. So any admin in the, in your Anthropic account can either at the individual level, the group level, or the org level as a whole, set a spend cap. The most restrictive spend cap always wins. So let's say you have a spend cap for a $100, and then they're all on a monthly basis. A $100 per month for one user, but you set an org spend cap of $50 per month. Once $50 in that given month is hit, all spend stops. Spend will not be able to or usage, I should say, will not come back until either the month resets or you increase the spend cap. So my recommendation to you all as you get started is set at minimum an org spend cap so you can budget this accordingly, and then identify your power users, see who are getting the most out of the the the tool, and you can use the compliance API to actually review what individual users are using. Obviously, this should be ran with actual, work use cases and not personal ones. But you can identify your power users and then give them a larger individual spend cap versus the folks on the team that, aren't using the tool as much. We can set a lower spend cap for them and also do discovery around, hey. Why aren't you getting the most out of this, etcetera? Great. Thanks for going through that, Drew Parker. The next bunch of questions we've received are really around international offerings and also additional partners that we're planning to release through our road map. So, Peter Nolan, do you wanna take a few minutes to talk about how we're thinking about this first wave of our launch and additional opportunities we have going forward? Absolutely. And so we absolutely have plans to expand the MCP ecosystem to incorporate a number of the tools, that many of the advisers are using today that that aren't on our our initial launch list. The way we go about selecting them is reverse inquiry. So we wanna hear from you. If you have recommendations on tools and partners with whom we should in include in the ecosystem and prioritize, we'd love to hear it. I think we saw a number of requests specifically for LPL and what their plans are with regard to Anthropic. I would point everybody towards a resource that we'll post in the doc tab. We just did a podcast with them yesterday, where Matt and Yeti, Greg Gates, and I walked through their their AI strategy. So, so that's one, but we would absolutely love to hear from you all on, on what should come next. If you're an integration partner or if you'd like to become part of the ecosystem, we would also, welcome you reaching out. Please just just be gentle. Don't don't react by getting angry. You have missed the first wave. We're going to add many more partners moving forward and are excited to speak to you. Excellent. We've had a number of questions around how to customize these tools for a specific firm, particularly around interpretation of compliance rules, particularly around firm specific workflows, particularly around firm specific branding and guidelines. So, Drew, I know you touched on this for a minute here, but I wanna spend a few minutes to share about how we're thinking about these plugins as an initial draft and how we can really help support people bring their own flavor to this offering. For sure. So there are a couple of ways that you can do this. I'd say if you're a platform or you're a large RAA, and you want a fully customized version of this plugin with each of these skills sort of, with layered firm context to your individual processes. You wanna input your brand guidelines to any of the actual outputs that are being created. As a firm, you can, toggle off any connector, plug in, or anything that's available to your end users. And so if you want, you can gate this and say, hey. We're gonna take this plug in to an initial pilot group of users, identify the actual customizations that we want to make, layer those onto our own version of the plug in, and then you can programmatically ship that updated version of the plug in to all of the users. You can either save it as a net new plug in. You could call it financial advisers dash your firm name or title it, whatever you want, and then you can push that down to all the individual users. Or if you prefer that advisers on a team customize the the skills themselves, that's where you can just turn the plug in on for everybody. And then individual users within their co work session can layer on their specific customizations. But yeah. So lots of different ways that you can do it, but that's those those are a couple. Great. We have another theme of questions around what is the best model to use for which workflow. Drew Parker, you touched on this early in the week. Do you wanna spend a few minutes on that? For sure. So I would recommend everybody start using this plugin with Sonnet. I think and I I can give some additional context on just, like, our model menu. So Haiku is our smallest model. It is our least powerful as well. It is great for very quick, very easy tasks that don't need a ton of thinking. Sonnet is gonna be your daily driver. It is an incredibly powerful model, but it is far less latent than Opus and certainly Fable. I would say for a majority of the tasks that you run, use Sonnet. Claude will return answers to you quickest, and it is, again, incredibly powerful. For harder tasks, you can boost up to Opus if you needed in a in a some additional thinking or inference. It will also be more expensive. And then for your absolute hardest tasks, that's where you can use Fable, even more powerful and more expensive than Opus. And I think as you think about what is the actual best model to use for a given task, you can just ask Claude, and Claude will be helpful and honest to you. You You can say, Claude, here's what I'm looking to do. What would be the best model to use here? And Claude will answer. But, yeah, I would say definitely daily driver, Sonnet, boost up if needed. And then firm owners and admins, that's where you can just set the model defaults accordingly for each of your users and also limit the the model menu if you'd like to as well. That's great. Another theme I think we should dig into is is trust. You know, how are we protecting client data and why the enterprise plan here is a fit? So, Peter, do you wanna take a few minutes to explain how this works for this industry and how we've thought about bringing this to the RIA Absolutely. Thank you, Belinda. This is a critical initially. question, everybody. So thank you for for, one, asking it and, two, uploading it. So how do we protect client data and what guardrails do the connectors have around that? I would point to three things primarily. First is, Claude works within each person's existing logins and permissions. So it only sees what that individual user can see. So as you connect to CRM, custodian, performance reporting, etcetera, the user is using their existing login so that Claude gets exactly that person's access and nothing more. There's no shared superuser account. If if an associate can't open a household in, you know, Wealthbox or Schwab, for example, Claude can't see it for them either. And firm's admins will choose which connectors are enabled. So a short version would be, Claude inherits your permissions. It never gets more access than you already have. So that's one. Two, nothing goes to a client or into your CRM until you approve it. A couple of questions popped up that we only have read access to Schwab. That's correct. We're starting with read only access into Schwab. We're starting very conservatively. We do have right access into a couple of our MCP partners, and we very carefully architected this such that you will have to approve human in the loop before it makes any updates to your core systems. Three, and probably most importantly, is we do not train on your data by default on team and enterprise plans. So something that I can't underscore enough, something that I would want everybody to take away as a primary conclusion from this call is we do not train on your data. And in our enterprise plans, we have audit logs. We work very closely with heavily regulated firms to design a compliance API and audit capabilities that allow you to satisfy your regulatory responsibilities. If you have any questions, I would point you and your chief compliance officers or CSOs for that matter, if your firm is big enough, to Anthropic. It's our trust center. From that trust center, you can pull down our SOC two or ISO certifications, our data privacy addendum, and anything else that you would otherwise need to demonstrate due diligence as as part of the SEC or or FINRA regulations. So hopefully, that answers the questions. If there are any follows, please don't be shy. This This is a critically important part of adoption and what we really wanna solve for you. That's great. Another theme that I'm picking up on is explaining a little deeper about the deterministic and nondeterministic nature of a skill. So, Peter, do you wanna go through how we work with our partners to ensure what's coming in from the MCP connectors is deterministic and accurate and how people can trust the output of some of these questions that they're asking for? Yeah. This is another critical question, and thank you for asking it. So, LLMs and AI are very valuable and helpful because they can reason, but they're probabilistic in nature. Meaning, you can ask it the same thing twice, and it may come back with a slightly different response. And so how can you, you know, effectively or adequately rely on something that isn't necessarily gonna be exact every single time is a very legitimate question coming from a heavily regulated firm. At the same time, you're using technology today that you do trust with those very important deterministic calculations that has the margin of error that you've been, putting at the at front and center in your business for many years. And so that's why we built this ecosystem around those tools. Claude fits on top of the tools that are in your stack today, running deterministic calculations, that you're already showing to clients. So all Claude is doing in this case is orchestrating. If you have something that you wanna show a client, the heavily regulated deliverable probably gets run in the MCP partner and then pulled from that partner by Claude and delivered to you. So that's a little bit about how we're navigating it today. We're making sure that the deterministic accounts that your business relies upon stay in the tools that you're using today, and that Claude is merely orchestrating across them to make workflows end to end, automated for you. Does that make sense? Or Belinda, anything you would add to that? Yeah. I would encourage people to, to start to use the tools and understand them. I think as people are starting to learn to use these capabilities, you know, trust and verify is a really important part of understanding how the technology works. So going through the first few queries, understanding where the data sources are coming, trusting and verifying, and then also working with our partners to understand what data is available through their MCP and how that's gonna show up in your query is an important way that you can build trust with the output that you're generating from your queries in Claude. Okay. I think there's a couple of other questions here around how the credits work. So we're gonna get to that in a second, as we close out. But I think in the in the meantime, you know, we very much appreciated all the engagement and questions. I think we will have a a roadmap that will include additional partners and international regions coming. But as you can tell right now, we're very much focused on the America's partners that we've we have listed today. So I think with that, Peter, why don't we, flip over into our next steps and wrap up our webinar? That sounds great. And, Drew Parker, feel free to chime in here as well. But, folks, we have a number of wonderful partners who are each hosting their own webinars, who have their own collateral, and we're able to educate each of you on what exactly the MCP and the Claude is capable of doing today and what their plans are for those MCPs moving forward. So a lot of the questions that were asked in the chat can be answered better by our partners. Please feel free to reach out to them. I'll quickly summarize them. Wealthbox, Orion, Adipar, BlackRock Investment, iCapital, Schwab, SS and C, Vanguard, wealth.com, Zox. And then we have our own connectors library with a lot of the core tech that you saw on the other slides and with our market data partners. If you wanna see more from these partners, give them give us the feedback. We wanna create this ecosystem to to be much more deep. We see this as a phase one. There will be many more phases to come, and we can't wait to bring you more capabilities through Claude. So thank you everybody for tuning in today. We're really excited about it. Drew, did you wanna touch on anything else before we wrap? No. I think this is perfect. Alright. Thanks, Thanks. very much.