Name: Claude for Nonprofits: Moving Your Workflow Beyond Chat
Uploaded: 2026-05-06T18:32:07.204Z
Duration: 1 h 16 s
Description: Claude for Nonprofits: Moving Your Workflow Beyond Chat

Transcript for "Claude for Nonprofits: Moving Your Workflow Beyond Chat": Hello, everyone. Thank you so much for joining us for our Cloud for Nonprofits webinar today. I'm Melissa, and I lead the partner success team at on beneficial deployments. Since our cloud for nonprofits launched, I've been working alongside organizations to help accelerate mission critical work and identify opportunities to leverage cloud to improve productivity so that organizations can focus on what matters most, connecting to their communities and accelerating their missions. Having worked in the nonprofit in education sectors my entire career, I can't even count the number of times I found myself wishing I could just do more. Honestly, time was the biggest barrier. By the time I finally sifted through program data to really understand point pain points and gaps, it was already time for a new cohort to launch, leaving little room for the ideation in the improvements where I actually wanted to focus. Similarly, the times I enjoyed the most were the times I were in front of beneficiaries or educators, really interacting face to face. And honestly, this is where I think I brought the most value. The realities are that I actually spent most of my time behind a computer. If any of this resonates, this is for you. One of the things I've found working with hundreds of nonprofit organizations is that there's huge value and benefit to offloading some of the more manual work to Claude to allow me to focus on what matters most and be more proactive in the way we support organizations. And all of this starts with taking your first step beyond chat. So this is gonna be our focus for today. I'm joined by two of my colleagues, Alex Casavan and Alisa Feng. They're also on the beneficial deployments teams. They'll be here to answer questions as we go through, so please go ahead and start asking those questions. And, of course, we will save time at the end. Just a few quick housekeeping notes before we begin. Afraid you're gonna miss something in the webinar? Do not fear. We will send out a recording that will be distributed within twenty four hours. Again, if you have questions, please ask. We have some time throughout the session to answer some of these, and we'll also have that dedicated q and a at the end. And lastly, we love feedback. Please go ahead and rate this webinar by selecting the survey widget at the bottom of the screen at any point during this webinar. So with that, let's go ahead and dive on in to the agenda. What I wanna focus on today is really being hands on. I wanna focus on building. Therefore, I'm gonna spend the majority of the time in cloud itself. I'm gonna quickly distinguish some of the features that we'll be going through today and introduce you to chat to co work and skills in a few more. I also wanna introduce the chat to co work maturity arc, and then we're gonna dive into the demo and as I mentioned to the q and a. Before diving into this content, I wanna show you an output that we're gonna build towards together today. Here's an impact report reporting six months status of a grant. This is the kind of document that at most organizations takes weeks. Someone pulls the spend data from finance. Someone else chases program staff for outcome numbers. Another person is writing the narrative. It then goes to the designer and then likely an ED for review. By the time it reaches the funder, the data is about six weeks stale and has taken a small workforce to complete. The promise of today's session is a tighter feedback loop, not AI writing the report for you, but the distance between something happened in your program and someone who cares about it being able to see this and being able to act on it. This window gets dramatically shorter. So over about the next fifty five minutes, we're gonna produce this exact document along with many other artifacts such as a board deck to go along with it, starting from a blank chat window. And as the name of the webinar suggests, we're looking to move beyond chat, but let's start there. Gen AI started showing up as a chat in the workplace a couple years ago. It's been widely adopted by just doing that, chatting with it, going back and forth, asking questions, getting summaries, writing drafts. It has proven to be a very productive starting place, but we also know that it has its limits. Last year, AI took a huge turn in being able to action things on our behalf. This is particularly with Cloud Code. You might have heard of agents, which is basically AI that has memory, tools, and can action on things autonomously. When we released Cloud Code, this was really for developers and for code. It was largely focused on building stuff, testing stuff, and deploying stuff. Many engineers now outsource all of their code writing to Cloud Code. Again, very productive, but we're also focusing on a particular audience here. So CoWork is that next phase. It's bringing that agentic power beyond engineering. It does the work for you. The agent you can access through your desktop handles those complex multi step tasks via cloud, and this is where we're gonna dive in today. We're gonna be thinking about how we move through this maturity arc. We're going to look at how projects and connectors can give you better answers with added context. Cowork can really serve as that inflection point that will allow you to save hours per week and carry out tasks on your behalf. We'll also look at how skills can help standardize across both tasks and teams. So now we're gonna jump into that demo. While I set this up, I'm gonna look to my colleague, Alisa, to hop on and answer some of the early questions that have come in. Awesome. Thank you so much, to everyone who has joined us so far and are already utilizing that Q and A feature here. So one of the first questions I saw here is, that some of the content is really helpful. We wanna make sure there's a way to share it with colleagues who might not be able to attend this webinar. So, yes, this recording will be sent to you within twenty four hours and from there, you can share it with your colleagues. Same with other webinars across the website that you might find if you register for anything and can't make it live, you can always share that recording. Another great question I, saw in the chat that is relevant to our content today is how do I collaborate with teammates in Claude? So folks who might be familiar with chatting with Claude and starting those one off conversations so far, this will be a great extension of how to do that. Today, we'll cover projects, which is where you can create that collaborative workspace to, share that same context and instructions across your team. So those were some really good, questions that came up just in our first few moments here, but I'll pass it back to you, Melissa. Awesome. Thank you. So as we head into the demo, we're gonna ground in a fictional organization, Compass Community Partners. Compass Community Partners helps under resourced communities build their own pathways to health learning and economic stability. They work across a mix of US and international communities through three integrated program areas, community health, education access, and economic resilience. For today's purposes, we are going to play the persona of Jordan Reyes, who is the director of development and communications at Compass. Jordan writes grants, manages reporting, briefs the board, and handles external comms. As you might relate, Jordan is overstretched and wearing many, many hats. So Jordan has found a new grant opportunity, the Hollister Family Foundation. It just they just opened their spring cycle. They fund community health and education, typically grants between 75,000 and $200,000. They require a two page letter of inquiry before even being invited for that full proposal. It's due in nine days, so Jordan needs to get to work and get that LOI out the door. So we're gonna go ahead and start in the main screen, and we're just gonna start with a general prompt here. I'm gonna get this started. And the first thing I wanna point out is notice how much I had to type just to get started here. Let me go ahead and show more so you can see that full prompt. I've provided context on my role, the org description, program stats, the budget, the funder's priorities. I'm doing Claude's homework for it. I've typed in probably about 80% of this. I'll type in the same content the next time I have to do an LOI. Right? 80% of this is probably the same every time I'm doing this task. It's pretty frequent and recurring. You'll also note at the bottom, I put in a note where it says not invest inventing or estimating metrics. The LLM itself is intended to do just that. It's content completion. Right? So what I'm doing here is setting guardrails on potential hallucination and ensuring Claude isn't filling the gaps with plausible sounding information. This is one of the key gaps we typically see. People are judging Claude truly on that first draft, right from that first cold chat. This isn't Claude's ceiling. This is really Claude without that appropriate context. So as you can see here, our first output has come up. So scrolling through this, I can see that it's covering the key information we're asking for, but it's pretty generic. Right? Compass Community Partners respectfully submits their LOI. I've got some estimated outcomes. You're seeing that 31% that I pulled in. The rest of those outcomes are quite generic. If I were to scroll down, right, respectfully, I've got some work to fill out here and make sure my name is put in. This could read like it comes from any organization. So now it's time for that iteration. And I'm gonna go ahead and ask Claude to adjust this a little bit. So I'm giving Claude feedback here. Right? This reads like it could be from any organization. I'm asking Claude to rewrite that opening paragraph so that very first sentence is more of an image from our field. Right? Something that is more authentic to us as an organization. I've also provided, right, additional context on language that is appropriate for us as an organization. I specifically called out language I want to avoid. Ensuring that we're communicating our constituents, and in this case, right, our community health navigators in the most appropriate way and again, in a way that resonates deeply with us. So Claude is now going ahead and it's adjusting that. It's taken that feedback. You're seeing here, it's updated this first paragraph just as I have I had asked for that field image. It's now giving me that information. It's updated some of the other information in here as well, particularly calling out, right, and making sure community health navigators is really clear language in there. Here's what most people miss. I just gave Claude a really useful piece of information and direction. And in a normal chat, this actually evaporates the moment I close the window. So watch this instead. I'm now gonna go ahead and ask Claude to add this to my memory. I want Claude to remember who I am, the role I have in what I do. I don't wanna have to repeat this every time I'm coming in here to do any sort of output, not just my LOIs. I've also put my writing rule in here. Right? In any funder facing documents, I wanna make sure that the appropriate language is used, and I'm providing clear direction on what to avoid using. So as you can see, Claude confirms that it's been saved to memory. It takes a minute to sync, but I do wanna show you where you will be able to see this. So if you go into your bottom left hand corner and go into your settings, under capabilities, you'll be able to see your memory right here. So right now, it says no memory yet, again, because it takes it takes a moment to sync. But when you do, this is something you'll be able to come back and check. You will be able to validate that it's here, and you'll also be able to edit and augment it as you need. So at any time, you can come back and check this. I also wanna point out above this search and reference chats. This is gonna allow Claude to do just that. Right? Claude can point to and pull information from relevant chats. But with memory, you're guaranteeing that the critical information is being used. Now this preference is gonna travel with me. A new chat tomorrow, a different funder. Claude already knows that I'm Jordan, knows what I do, and knows those words to avoid. This is not something I'm going to have to repeat again. Memory is really how your individual judgment compounds the same way a a project and how your team's context compounds. So the bottom line, when you give Claude a correction that you'd give again, save it to memory. You can say don't do x, remember y. Don't treat this as a one off. Putting it in here is going to ensure that this is repetitive and becomes part of that practice that you're going through. So now that we've looked at memory, we're going to jump into projects. I'll talk through the value of a project in a moment and we'll be able to demonstrate that. First, I want to show you how to get the project set up. So we're gonna start with project, Compass grants. So I'm gonna go ahead and create a new project. I can put in a description here if I'd like, and then I'm going to go ahead and create that project. The first thing I'm going to do within this project is add project knowledge. Let me explain what project knowledge is. Project knowledge is just as it sounds. It is giving Claude information in context so that it can leverage this throughout the entirety of the project. A project should be considered that persistent workspace or collaborative space where you and your colleagues can work together. In this case, it's really how you're leveraging information across a broader set of organizations, or in this case, right, different grants, not just Hollister grant that we're focused on. And one of the key pieces is you can actually go in here and see your projects at any given time. So again, if I were to click back into the grant that was just developed, this is where you're seeing those files. For the purpose of this demo, I'm gonna go ahead and add some, and I'm just gonna simply upload from my device. So I'm gonna go into my desktop here, go into my demo files, and I am going to add some files here. And as you can see, right, I've started with some files and I could actually continue to build on this. So if I wanted to go in and add more, right, I can go in here and continue to put in additional information. So if I wanna run a second set, I can look here and make sure I'm including, for example, our boilerplate, our program one pager, a voice guide, past successful LOIs, and get this all loaded into my files. This is now context that I'm giving to Claude that Claude can use and leverage as we move forward. I also wanna show project instructions. Projects instructions, similar to the concept of memory, is something that is going to persist throughout the entire project. So here's where I'm giving a little bit more context. Right? You're supporting this team. When drafting right, funder facing documents, ensure that the outcome summary is being leveraged. This is something that we'll do every time now that this project is being used. So let's go ahead and prompt it here. I'm gonna ask for the same thing I asked in that previous chat. I'm just doing it a little bit differently. You can see already that Claude is starting to read the files. Right? It's leveraging that project knowledge, those files that it knows it needs to use in order to give me the best output. So I've asked for that two page LOI for Hollister Family Foundation. I've done the same ask as before, but now I've done it in 40 words instead of 200. I didn't have to remember the preventative care number or our voice rules or what worked in our last successful LOI. The project remembers. And because memory is also active, the words to avoid list that we enforced previously is also going to be added to this context here. This work, again, compounds across people. Anyone on the team that I've shared this with can then get that same quality draft because the context is living here and not in my head. I wanna call out one really powerful piece. I added three past grant examples to those files. LLM is pattern recognition at its core, and it's really good at this. So what I'm doing is simply providing it with that. I'm providing it with examples that serve as patterns that I want Claude to follow. You'll also notice that I chose to put in Hollister stated priorities. Right? I stated this in that last sentence here. As an alternative, I could have likely just linked their website or the grant page. What this would have done though is this would have given Claude the permission to discern what context was actually important. Because I'm the expert here, I'm able to discern this. Right? I'm able to understand what is relevant and then feed that most relevant information to Claude to really narrow that wedge and ensure I'm getting the best output. Okay. So now that this has opened up, right, we're seeing that it's opening up with that field image just as we wanted it to start with. It's giving me the problem. It's pulling multiple outcomes. If you remember that 31% that I had provided previously, right, it's including much more than that. It mirrors our priority language, and those banned words from memory are gone. You're also noticing here, right, that we are seeing our brand represented here. This is now giving me something much more sufficient to start off with. I mentioned before, right, creating an LOI is not a one off for Justin. This is something that is done frequently. So now I'm gonna show you how we can leverage the skill to really drive even greater efficiency here. So first, what is a skill? I want you to think about it as a recipe card. So if you've ever written a how to guide or a set of instructions, something as simple as how to create our weekly report or how to submit a contractor request, you've really already written a skill. Now you're just storing it where Claude can see it. I'm gonna go ahead and fire up that skill. Once we run the skill, I'm actually gonna walk you through how to create one so you'll also get that inside scoop. But first, I'm just going into my skills here. You're seeing this compass LOI one that I've already started. I have this turned off just so we can leverage that first part of the demo where we were just using chat and then project. So now we're layering on. Now we're building in our skill to this process. So now that I've turned on this skill, I'm gonna go back into that same project. What I'm gonna ask for now is I'm gonna ask Claude to use that skill to draft the LOI for Hollister Family Foundation. Very similar to before, Claude has all of those context files, so I have to give Claude very little information. You can see it already picked this up. It's reading that skill. What it's doing is it's reading through the instructions so it knows what to follow. You can see it's already called out. Right? It's going to that outcome survey. It's going to the data that I've referenced in the skill to ensure that it's pulling what is most relevant. So as this is going to start cooking, I'm gonna go ahead and walk you through what it looks like to actually create a skill now. So again, if you go to customize and then click on create a new skill, you're gonna have a few different options. I'm gonna go with right skill instructions. You could also create one with Claude in a more collaborative manner, or you can upload a skill. I wanna make sure you have this visual representation of uploading a skill as well because at the end of this session, we will be leaving you with a prepackaged skill that you all can leverage. Okay. So for the purpose of this, I'm just gonna do an LOI, writer, and I'm calling it demo since I already have that other one. So whenever drafting a letter of inquiry concept note, right, I'm giving Claude the kind of when to use in my description. This is going to ensure Claude understands when to pull it. Then I'm gonna go ahead and put in my instructions. Right? I'm listing out here very clearly what are the sections of that LOI that need to be included and what information should be used. I've been very, very specific even on what that output should look like. Noting when I want a paragraph, right, the number of sentences I potentially want. If I wanted tables here or bullets, this is also pieces of the output that would be really helpful to note to note. I also wanna call out this. Right? Always end with a section titled, verify before sending. This is something I'm typically putting in when I'm developing external communications Because what it's doing is ensuring that it's calling out sources that I need to go back and verify. Right? Making sure that I'm playing my role as human in the loop to ensure that I'm discerning this before determining how I want to leverage this particular artifact. And then I can go ahead and click create. And it's as easy as that. The name, the description, and the instructions. Once this is developed and saved, I can come back to it anytime. So if you see an output that is missing context consistently, that might actually just be a skills issue. So if you forgot a step, no problem. You can always go back to this and edit it anytime. Okay. So now that we've developed our skill, I wanna toggle back to the project and open up what we've done again with our skill. So one of the first things I wanna point out here is this verify before sending. Just as I had put in that skill, right, it's flagging items that need my review. If I were to scroll through this, we're seeing that it has the logo. Right? It's giving me more of the information here. It has a much robust it has that robust set of metrics, and then it has to verify before sending and the word count just like I've asked for on every LOI. So just to kind of take us through where where we've gone, we've gone to see the same output going through the process of, projects, chat projects, and now leveraging skills. Right? Really being able to think about how we're building here. We can continue to compound here. Right? I wanna start and now highlight connectors, which is another really powerful way. Everything so far has really relied on what Jordan already knew about Hollister. But what if Claude could go and search or find the next Hollister while it's on this topic? With the help of connectors, we can. So similar to where I go for skills, I'm gonna go into my apps and I'm going to connect a few of the connectors. So I'm gonna add Candid. I am going to add, granted. Right? These are a few of our connectors that I'm really excited about, and I know I will leverage consistently. Also gonna go in here and add Kendora. Great. So I've added a few connectors and now let me go back to my project and you can see here. Right? I can see that these connectors have been added. So now what I'm gonna do is I'm gonna ask Claude in a new chat. Based on that LOI, can you use the granted connector to identify five other potential funders I should be applying for? You notice here that because I've already done that LOI and I have on the ability to search relevant chats, I don't have to go ahead and even pull that previous LOI. Right? Claude is able to pull that context from here. This is really what should feel like a shift in the type of work. I'm not feeding Claude information anymore. Claude is feeding me information from a live source, and then I can determine how to act on it. So it's the same writing, just different doors into your data. Once I receive these insights, it's then really up to me to determine which direction I wanna go in. Right? I may continue to chat with Claude back and forth to understand more of the details, ask questions, brainstorm top priorities or the best contenders. Alternatively, I may know which direction I want to go in. So if this were the case, I could actually fire up that skill again, and I could get that LOI out the door pretty seamlessly. What that's gonna allow me to do, again, is just continue to compound that work. I'm working directly in cloud, and I'm able to pull that information. And then I'm able to think about how to leverage that. Because this is taking a moment to load, I'm gonna jump over quickly to show you the example I ran earlier. Right? So same prompt I had done here, And it went ahead just as I just as I had prompted, and it pulled these different potential opportunities. Right? So just as I mentioned, here are now those five opportunities that I could start with and use them as a starting place. Instead of having to even go outside of here and think about doing the work elsewhere, I'm able to pull that context in and to continue to build off of the work that we had just shown. So now because this is a webinar, we're gonna fast forward a little bit. We are going to think about the LOI is out the door. And six months later, we've got the grant. A $175,000, eighteen months for the community health navigator expansion. Per the grant agreement, Compass owes Hollister amid grant reporting, covering activities, spend against budget, and early outcomes. Jordan also needs to brief the board next Thursday and wants to share some of this critical information with the community as well. So we have three audiences with one set of facts. We're gonna focus on ensuring we are not duplicating work here. We're doing this differently than the LOI. Instead of going to show you what it looks like when that ladder is we're gonna show you what it looks like when that ladder is already built. Right? A project that's been collecting grant activity all along, a scheduled task that is keeping it fresh, and Claude doing the analysis, the design in the deck. So I'm gonna switch over to CoWork for this. And I'm going to jump into a project that has been going for this grant. Right? So now we're looking at Hollister Grant Reporting. Very similar to how you saw in the other project setup, you'll notice that I have a set of files here. This is the context for my project. You'll also see that some of these files say live. These sync with real data that we're pulling to ensure accuracy without having to touch it. This data in the project didn't get here by Jordan uploading it. It's been arriving on its own through scheduled tasks. So I'm gonna go ahead and show you an existing task. As you can see here, there's a scheduled task. It says weekly pulse. If I scroll up, I can click into the run task here to give you the description. So every Monday, this pulls the latest health navigator activity log and that spend tracker from Drive, and it's pulling it into the Hollister grant reporting project. It has specifics on what it needs to compare and then what that output should look like. So as you're seeing here, this is an example of what it's pulled on that time cadence that I've provided. So if you think about the power here, this is something that should be running since that grant is started. Right? You're setting it up once. I didn't have to build a dashboard. I don't have to ask finance to send me that report. I'm describing what I want and when I want it, and Claude has been checking for me. This is really that tightest version of the feedback loop. Right? The distance between thinking about overspending on some of this training material and Jordan knowing about it is immediate. So now let me show you how to go ahead and add one. I'm gonna go into scheduled. I'm gonna click on new task, and I'm gonna put in a story bank. And I'm gonna use this to collect stories. So what I'm doing here, I'm also gonna connect it to the project, is I want Claude to be able to pull in relevant anecdotes from outcome surveys so that as I'm developing materials, I can deliver the most up to date information. You'll see here I can also put frequency and adjust that. And And for the purpose of this, I'm just gonna go ahead and and save this. Claude confirms up here that the story bank task has been created. If I were to click in it, this is also something that I can go ahead and run now. Right? So it's running, it started, and I could even view this immediately outside of that scheduled task. That's it. Now I have a story bank that fills itself. So when a funder report is due, when a board wants a vignette, when comms need a social post, the raw material is already waiting. So now we're gonna jump back into that project. Let's go ahead and actually run some of the analysis. Right? This is this is the actual ask here. So what what we mentioned is the first part is that we need to run analysis that is going to be delivered to Hollister. So let me go ahead and put in that prompt. And you'll notice here, I'm not uploading anything. I'm not defining what on track means. The grant agreement is in the project. So Claude knows the targets. The spend tracker is also in the project. Claude knows the budget. I'm just asking the questions that a good program officer would ask. Notice the risk section here. What I've included in the prompt, anything a smart program officer would ask about and what I should say before they do. Right? This is the move here. Don't just ask Claude to summarize your data. Ask Claude to interrogate it. Right? Think about some of those toughest stakeholders, what they will ask, thinking about defining those roles. This is also going to close the gap for what Claude needs to be focused on and give clarity to that audience that is most important to you. So now that this is running, thinking about what I'm gonna be looking for here. I'm really gonna be looking for that clean variance table. Right? Ensuring that it matches the grant agreement, which is validating that it was pulled from that files. Making sure it includes the outcomes and activities. I wanna ensure those targets are being pulled in from the context. This is not a piece of information that I typed in. And if I have to do a nine month, a twelve month, and then that eighteen month analysis, I don't wanna have to keep pulling in the same information. I'm able to just lever leverage it here. And then I wanna make sure that risk section is really surfacing what's critical. Right? I wanna ensure that it's pulling the most relevant information. So in the interest of time, we're gonna fast forward, and I'm gonna show you what that actually looks like as Claude is thinking here. This is what I had run earlier. As you're seeing here, right, this is giving me the full analysis just as I mentioned. I have that variance table with those explanations. It's looking at the targets. This is what was pulled from the grant agreement. I've got the data. And as we go further, you'll see I've got the risks here. So this is now that foundational piece of information that I can use to leverage many different types of communication. So let's think about how we can leverage it from that perspective. Now I wanna turn this into a community impact update. Right? I wanna think about how I can leverage this for my community, just as I mentioned at the beginning. So thinking about the same set of data, leveraging it in a different way. So I'm gonna go into my project and I'm gonna go ahead and ask here. I wanna turn this into a two page community impact report. This is really where, right, that Beyond chat stops being abstract. Claude isn't giving me text to paste into a design tool. It's going to be producing that designed artifact in the brand ready to print. Right? I'm asking for that finished PDF. I'm pointing it to brand colors, and it has that information because it's part of the files. And I want to ensure that it's pulled those stats, the charts, as well as now those quotes from the story bank that I run. This is something that is now a living, breathing artifact that I can use. So an important thing to keep in mind here, don't re prompt when the audience changes. All you need to think about is re aiming. Right? Tell Claude who it's for, what that person needs to walk away knowing. The project holds the substance. You're just changing that lens. So now we're gonna jump into that what that community impact analysis looks like. And because I wanna show you, right, we asked for a final PDF. I wanna be able to show you just that. So as you're seeing on the screen now, it is giving me just as I asked for it, that final PDF that I can use in Leverage. You'll see it's got the quote in there. This is pulled from our story bank. It has the outcomes that we wanted to highlight and is largely ready to go. This was done in that one prompt shot. As I mentioned before, in your role, right, there's likely going to be that iteration, but you can see this is again something we were able to do very quickly. So now the next step there. Right? We've talked about leveraging this for that funder narrative as well. Right? We're thinking about what is the next step. So we wanna create that version for Hollister. So with this, same thing. Right? I'm going through another versioning here. And I'm saying now the version for Hollister. I want them to take that analysis and write that two page from that two page and create something completely different. So here I'm specifying a few of the different outcomes. Right? The challenges and adaptation section, making sure that we're proactively addressing missing pieces, and mirroring the section headers from the reporting template in the grant agreement. So, again, I'm referencing what is critical, and it's gonna go ahead and pull that. Here's the example, if we fast forward again, of what it looks like. So similar to before, this is one where I can look at this directly in Claude here. I'd also, because it's a doc, could export it as I needed in leverage there. So with this, we're seeing the section headers just as I had asked for in that prompt. We're seeing the spend, the outcomes, those challenges and adaptations. All of this is consistent. If you recall from some of those other artifacts, right, staffing delays, the training materials, the overspend, these are all the same pieces that were called out consistently but are now being changed for audience. Right? These are not relevant. We're not relevant for those that community audience. So it was not included here. It's been pulled in here because we're giving Claude that relevant information. And again, very similar, verify before sending. So that I know at the bottom where human intervention is really needed and can focus my time. So now that we've gone through, we've taken the same p grant analysis in the same piece of content, and we've developed it for different audiences. We're now taking that last step that was critical for Jordan, and we're gonna develop a deck for the board. This is where I'm actually gonna go into Claude for design now. So you see that I switched screens here. I'm in our cloud cloud.ai design tool and wanna walk you through how I'm gonna go ahead and leverage this. So two things I wanna call out. First is this idea of creating a design system. I went ahead and created one earlier, but I do wanna show you how easy it is to do so. So I can go into design systems. I can select create. In here, I'm just gonna put I'm gonna put demo design for compass. I have options for what I need to put in here. For simplicity purposes, I'm just gonna go ahead and browse, and add in some pieces that are most relevant. So I'm just gonna put in that logo in brand kit. Right? This might not be all that your organization has. You can put as much as you want in here. But I'm starting here, and then this is what I need this is all I need to kind of get that started. I can then continue to generate. It notes here, it's gonna take about five minutes to generate that design system, which is why I did it in advance. But of course, I could go ahead and run that, and then you'd have your design system that you could even use as a starting place and continue to iterate on. So now we wanna build that deck. This is where I'm gonna go into slide deck and I'm gonna put here board report for Compass. You'll see it has my Compass design system that I pre built already in here. I'm gonna go ahead and create that. I'm gonna put in my prompt. So I'm here specifying what I'm looking for. I want those five slides. I'm gonna go ahead and attach as well specific files that I want to ensure are coming in here. So in my Compass grant project, right, I want to ensure that I'm pulling in relevant information. In this case, right, I might wanna be pulling in whether there's particular funder voice so I can go ahead and open that up. I might want to go in and go into the artifacts that I created for my project. So I'm going into the Hollister grant reporting, and I'm gonna pull in that six month analysis that we just ran. Again, thinking about leveraging those same pieces of information, and I'm gonna put in that community impact report. So I've uploaded relevant files or what I believe to be most relevant for that first draft. Same facts, new lens. Right? A community version led with the people. That funder version led with deliverables. This board version is going to lead with governance. Are we on track? What is at risk? Is there decisions that need to be made? And this is the thing I want you to hold on to. Right? All numbers in the three documents will agree. This is because they're coming from one analysis in one project. Nobody's going to open that board deck and get a different household reach figure than what's in the funder report. That consistency really is a quiet superpower of doing this in a project instead of three separate chats. This is going to give me that PowerPoint that I can use in leverage. So if I go back here, I'm gonna show you what that final output looks like. And here it is. Right? Same prompts. Here's the prompt you're seeing that I put in. Build that five slide PowerPoint deck. It's listed out what I want in those five slides, and I can now come in here and see this pretty polished version. This is a great starting point. It might not be that final version, but it definitely helped me get through that blank page problem of just staring at the PowerPoint. I now have something I can start to iterate on, and I can do that just by going back and forth with Claude here, or I can go directly here and just edit it. So that's it. The document I showed you forty five minutes ago was generated along with a community one pager, grant analysis, and now this board deck. We went from application to award to analysis to final report in the time it took to watch this webinar or attend live. Right? That's the feedback loop. So with that, I wanna jump into q and a, and then we're gonna end. I mentioned this at the beginning. I wanna come back and share something that you all will leave with. So I'm gonna go ahead and stop sharing. I am going to pull up some of the questions that are coming in. And we will get started. Okay. So a question that came in is, I'm just understanding the difference between a chat and a project. A so let's start with again that that chat. Right? This is that one off conversation. Hopefully, you saw this in the beginning of that arc. Right? Thinking about that context that could potentially disappear, where in a project, it's really that persistent workspace with shared instructions, files, and that memory. So that's what's going to give you the added context and really improve that final output. Another question came in around hallucination and how do I reduce hallucination. Right? One of the key things you probably saw me do in this deck is tell Claude not to invent figures. Right? You can also leverage, certain elements such as saying use, use a specific piece of research if you have it in the files or flagging when you need a source called out. If Claude is trying to put in a placeholder for a metric, have Claude put this out or pull it out in brackets. That verifying that I did at the end is another great way to do this. You can also ask Claude how confident they are in the response or ask Claude to ask follow-up questions to ensure they have the necessary context in order to move this forward. That's really going to help ensure you get the most out of it and give you greater context in that output. The key piece to keep in mind is the critical piece of human still being in the loop in really wanting to ensure that we can take advantage of that. Great question. So another question came in around the idea of skills and also thinking about how it's different from a project. A key piece to keep in mind here is that a skill can actually layer into a project. So for example, the project that I had focused on was for grant LOIs. I might have take take a different path. Right? If I had a project for maybe all of my, economic grants, one for all of my education grants and another one for all of my health grants. Those skills might actually be the same. That skill might be the same. Right? I might develop the LOI, consistently for those, but the context would be different. That would be a case where I could develop one skill, and I could actually pull it into all of those projects as we're going through. There was a question on how you can make projects org wide as well. So let me when you're in your project environment, there's two ways to do this. When you create the project, as you saw me create it, I created it in private. You could also create it public so that it is available for the entire organization. What you might also want to do is just share it with a dish with specific individuals. So once the project is created, you will see in the project space a spot where you can click share and then you can enter individual emails to share it with. One of the pieces to flag is that you will have the ability to give them view access and also edit access. The key difference here is the level of control. Right? So if you're giving them view access, that means that they cannot adjust the files or the project instructions. Right? They can interact with it and take advantage of the context there, but they're not going to be able to actually go and edit it. Versus if you give them edit access, they kind of have full control over it and can can do just that. So something to keep in mind as you're thinking about how to to go through this with your your organization. If you cannot see some of these features available that I went through, I would recommend that you check with your, with your organization. It is up to each organization to turn on features that you all are comfortable with. So we leave this control in the hand of organizations to ensure that it meets your internal safety and security policies. So I would encourage you to take a look, at your your policies and use that as a way to, make sure you're getting the most up to date information. Alright. I'm also getting questions just about and thanks so much for all of the interest. Questions about, signing up here. Right? Looking at the different plans that are offered for for nonprofits. So excited for for folks here who are, not claw for nonprofit partners. We will share this information out with the webinar as well, and make sure you have our landing page where you can find more information, in order to to sign up if that if that's of interest to you and your your organization. Okay. Another question, that keeps coming in pretty frequently is thinking about the best steps for nonprofits to onboard. Here's what I'm gonna start with based off of the based off of what I know working with the organizations I have. This is going to vary. There is not really a one clean run book because it depends on where your organization is at. There's a few things I think that are extremely critical as you're thinking about rolling something like Claude out. One is starting with your why. Right? Truly understanding why this is important to your organization and thinking about what your priorities are. I typically work with organizations whether they're working with whether they have OKRs or any specific goal model, and I start there. I take Claude out of the equation completely. Right? What is important to the organization? What is going to move their mission? What are their big bets? Then we start with how can Claude help there? That is what is going to get that early buy in and get that investment. The second piece is developing a plan that takes time. This is not something that's gonna happen overnight. Learning Claude, AI in general, is new. It's learning a new skill. Expect there to be a period of ramping up. We we typically talk about this as this j curve of research, Right? Where organizations might start off and they really have those early wins with chat. And then it's starting to grapple with where do I go from here. Right? And how do I really use this to accelerate my work? And that's where it's being intentional about what work you want Claude to take on versus what work you should still maintain. If I were to mention a few resources, and again, we can share more out after, after this webinar, it would be Anthropic Academy has an incredible number of resources. One I would really shout out is our AI fluency course. This is really going to help build that AI fluency, which we consider that durable skill. The last thing I would recommend in terms of onboarding is building your own internal champion network. Right? Finding those who are early adopters and who really wanna get on board so that they can start to engage. Right? They can start to be the advocates to really share across the organization what's working and what's not. And my last tip would be honest and transparent in your organization. I often ask leaders to lean in and talk about the challenges that they're having with Claude. You're not always going to get that perfect artifact on the first try, the second try, the third try. Right? Share those wins, but also share those fails. Those fails are what we're learning from, and that's really what's going to help you all as an organization propel forward. Another question. Can projects have subprojects or is it possible to create a set of institutional organizational documents, that are common across projects? Fantastic question. So the the first answer is there are not the ability to create sub projects right now. But what I think you're getting at, if I'm understanding the question correctly, is you wanna make sure there is informational documentation that is pulled across multiple projects. That is a perfect use case for skills. So some of some skills you might wanna think about is if you leverage brand guidelines. Right? Being able to have brand guidelines would be something that you might wanna be pulling across multiple projects, and you can set that across the organizational level. Similarly, if there's any compliance, for example, that should be followed, you can have a skill on compliance. Think about what is going to start coming up in projects consistently across that. And those might be areas where you actually wanna start developing some of those skills. Alright. I think I'm gonna have time for about two more questions before ending here. So question on protecting data. So really, really great question here. I'm seeing a few a few around here. So one thing I wanna be very clear on, if you are in our Cloud for enterprise environment, we do not train on your data. So the data that exists, your inputs and outputs in your environment, that is solely for your purposes. With that said, you should absolutely follow your organization's policies and ensure that those are being being adhered to. Okay. Another question here, and this will be the last one. Will there be more sessions like this to attend? And where is the best place to check? Fantastic question. Yes. There will be more sessions, we hope. We this would be a a reminder to please go ahead and fill out that, fill out the feedback. We will have more sessions. We want you to help us shape this. Please tell us where you want us to go deeper, what topics you want us to focus on, and we will send that out. So similarly, you likely got an email in order to participate in this webinar. We will go ahead and make sure that we're getting that information out as well, and hope you attend future webinars. I know there were so there were so many questions that are coming in. We did not get to all of them. We did not I will be transparent. We did not even get close to getting to all of them. We tried to cover what was most top of mind. I think this is a perfect place for my team to actually take back all of these questions and make sure we are building a more robust FAQ, and we'll deliver that on our website. So I don't have the timing for that, but that is something I promise I will do as an output of this webinar and ensure that we find the best way to reach you all with that. So with that information now, I wanna make sure we leave you with something. Right? We just went through a significant amount of information that you sat through, and I thank you so much for being such an engaged audience. So to end on, I wanna ensure I can quickly give you a leave behind. So as mentioned, you are going to receive an email within twenty four hours, and it is going to give you the recording along with some other documentation. We are going to give you a skill that you can populate just as I showed you before in your environment. You can go ahead and upload it. And you'll as soon as you upload it, you can tell Claude, help me figure out where Claude can help my organization. This is going to be your personal Claude assistant. We've built it so cloud will collect information from you about your particular work, and then it's going to leave you with a plan. That plan is going to include copy and paste prompts you can start with, each leveraging some of the features that we talked about. So if you have any constraints, make sure you let Claude know. And then it's also going to include links to the artifacts that support this. We have a use case library that we keep up to date and are continuing to build so that we can meet you all where you're at and ensure we're giving you recommendations that are most, that resonate best with you all. So know that more is coming here. And because I did see in the chat, some of you do not have skills turned on, we also wanna leave you with something. So there is going to be a project alternative to roadmaps. What this is going to be is it is going to be a set of project instructions and how it works. So we will give you the prompt in the project instructions that you just need to put in that project in order to get started. Once you have this, you will be able to take everything you need and use that as a starting point. If I can leave you with one last piece of advice, given again all of the questions that have come in that we will, we will work on addressing, but will take some time. When in doubt, ask Claude. Right? When you're stuck or when you're not sure where to go, Claude will likely be able to give you an answer. So I'm gonna take one of the examples that I answered. Right? I talked about hallucination and ways to avoid it. This is something that comes up very commonly. You can actually go ahead, ask Claude this question, and Claude will give you feedback. Claude is designed to be helpful, harmless, and honest. Right? I encourage you to lean into that and leverage Claude in order to best help you. So with that, we'd love to thank you for taking the afternoon to be here with us. We've really enjoyed, enjoyed being able to share more about this and hope you found this beneficial. Have a good rest of your day.