Community IT Innovators Nonprofit Technology Topics

Nonprofit AI: Claude for Small Business, ChatGPT Update

• Community IT Innovators • Season 7 • Episode 38

Share episode

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 24:35

Carolyn Woodard covers two recent AI product updates and a thought-provoking question about what it means to use AI tools more personally: a new Claude for Small Business plugin that connects AI to the tools your nonprofit already uses, a ChatGPT model update that changes the default experience for anyone on your staff using the free tier, and an article from nonprofit AI trainer Tim Lockie that may challenge how you think about sharing context with AI.

Anthropic launched Claude for Small Business as a plugin inside Claude Cowork, their agentic work environment. The plugin connects Claude directly to tools like QuickBooks, PayPal, HubSpot, Canva, DocuSign, Google Workspace, and Microsoft 365 through pre-built workflows for tasks like payroll, month-end close, and invoicing. Every action requires human approval before it executes. Nonprofits with a paid Claude plan already have access but need to make the connections in Cowork. The Claude for Nonprofits discount brings the Teams plan to $8 per user per month for qualifying 501(c)(3) organizations. A free AI Fluency for Small Business course is also included.

OpenAI updated ChatGPT's default model to GPT-5.5 Instant in early May, rolling it out to all users including the free tier. The big change: the model now draws on past conversations, uploaded files, and connected accounts like Gmail to personalize responses. If your staff are using the free version of ChatGPT, their default experience just changed, and that matters for what your organization's data governance policy says about which tools and tiers are appropriate.

Carolyn closes with Tim Lockie's recent piece "Humans Are The Loop," about building a private Claude project he uses as a personal thinking partner. He fed it his neuropsych evaluation, DISC profile, and StrengthsFinder results, and uses it to surface the patterns he is most likely to miss under pressure. This approach is in genuine tension with the data caution that guides most of our AI governance guidance, and Carolyn is still sitting with it. Worth a few minutes of your own reflection.

Resources Mentioned:

New every Tuesday.

_______________________________
Start a conversation :)

Thanks for listening. 


Carolyn Woodard

Hello and welcome to the Community IT Innovators Technology Topics podcast, Nonprofit AI Midweek Check-in. My name is Carolyn Woodard. I'm your host, and I am not an AI for nonprofits expert. No one is at this point. If someone tells you they are, they might not be telling you the exact truth. But uh we are

Carolyn Woodard

We meet here every Tuesday to learn a little bit of news, a little bit of resources, um, some ideas going forward and how you and your nonprofit can utilize AI tools effectively, securely, and you know, kind of things you need to know about AI as a nonprofit.

Carolyn Woodard

So today I wanted to open with a quick uh there's breaking news this week that's worth mentioning. Uh, if you've been following, it's in the news news uh this trial between OpenAI and Sam Altman and Elon Musk, which was just uh decided uh yesterday. As I'm recording this Tuesday morning, the uh jury came back, I think they deliberated for like two hours or something like that and found for open AI. The judge accepted that. I think Musk's team said they're gonna appeal.

Carolyn Woodard

Interesting case, I think for me, I was interested in it because, you know, for nonprofits, the core question about this trial was the ethical one of they were arguing about open AI was founded as a nonprofit, and then it created a for-profit out of the nonprofit. And uh Musk was arguing that he felt misled by that whole uh process. And so, you know, I just found it interesting that this giant tech case uh really hinged around like a nonprofit versus a for-profit.

Carolyn Woodard

And that question kind of ethical can you take an organization that was built on a charitable mission uh and restructure it for for-profit? Can you spin out a for-profit, part of a nonprofit mission? And this case, the jury didn't decide about that question. They decided on some technical grounds that um the lawsuit hadn't started within the Statute of Limitations. So they weren't able to rule on the ideas, they just ruled on the technicality.

Carolyn Woodard

So following that trial, even just a little bit, there's some good uh reporting on it in Wired. That I don't know, just this, like I said, the arguing over the nonprofit, the arguing over the leadership, the ethical uh decisions that leadership are making about whether AI is going to be good for the world and should be shared with everyone, if it confirms your skepticism over AI companies in general, that they're only out for profits. Uh, anyway, I I wanted to wait for some more analysis to come out before we really talk about it, but I wanted to just mention that because it did just happen.

Carolyn Woodard

I also wanted to talk about there's a new aspect of Claude for nonprofits. Uh, Anthropic, of course, is the company that owns and markets Claude. And I talked earlier on this podcast a few weeks ago about Claude for Nonprofits. They have a deep nonprofit discount for their paid tiers.

Carolyn Woodard

And they launched uh last week the Claude for Small Business plugin, uh, which sounds super interesting and maybe very valuable to your nonprofit. So they have created this plugin. It's available on any of the paid tiers of Claude, so Pro and Teams and above.

Carolyn Woodard

And what they did was they built in agentic workflow inside Claude Co-Work. And these workflows can connect Claude directly to the tools that you may be using that small businesses use and a lot of nonprofits use too. So, for example, QuickBooks, PayPal, HubSpot, Canva, DocuSign, Google Workspace, Microsoft 365, and some other uh some other tools, common tools that you may be using for you know your payroll or your HR or other tools that you might be using.

Carolyn Woodard

And since a lot of nonprofits use these tools also, you know, this just seems like it's going to again supercharge what Claude uh can do. And

Carolyn Woodard

It really shows Anthropic and Claude paying a lot of attention to the their market, small businesses, small nonprofits where you literally don't have enough hands, you don't have enough capacity to do things that you would like to be able to do. And so adding in these AI tools gives you that extra hand to, you know, make connections, do analysis, uh understand what your data is telling you.

Carolyn Woodard

It shows Anthropic really listening to maybe some of the complaints. I mean, I've had the complaints, I'm sure you have too, of trying to use an AI tool, even within Microsoft, to look at something in another part of Microsoft or wanting to connect to tools like QuickBooks that you're using and you know, another tool that you might be using to have better analysis and but still have the security and have those connections that doesn't take a coder to make those connections for you.

Carolyn Woodard

So, for example, if you use QuickBooks and PayPal, uh Claude can now help you with invoicing, payroll, month-end closing. Um, it can do analysis for you. So that might impact, you know, if you have someone on staff who, if they had this kind of assistance, could do simple bookkeeping for you or, you know, some of your financial stuff, then maybe you don't need to be hiring a consultant or a part-time CFO who's doing that for you. So it's really gonna give you more capacity and maybe change the way you're looking at some of these questions about what you pay for, who you pay to do it.

Carolyn Woodard

If you have Canva, of course, you can use Claude to connect to Canva and take your ideas to designed materials. You don't have to be a graphic designer. Of course, Canva already allows you to do graphic design without being a graphic designer. But um, you know, Claude is going to be able to connect that right up.

Carolyn Woodard

You know, if you have a document that you want designed, um, Claude can help you with that. And uh DocuSign, of course, contracts and agreements, and then HubSpot, Google Workspace, and Microsoft 365 can connect up communications, project work, collaboration, all of those things. Your files.

Carolyn Woodard

Of course, we talked before about being careful about files and access and permissions. So the cool thing about this, of course, Cowork within Claude is their kind of coding project area. If you have a paid version of Claude, it's a tab that you can see.

Carolyn Woodard

We've been talking about agentic AI a lot of, you know, it's out there. People are talking about agent AI as the step past generative AI where you can actually have the AI agent do a thing, and you can connect AI agents up in a row so that they each do part of the thing, and then you get the finished piece and how keeping the human in the loop when you're doing that is really important.

Carolyn Woodard

Not having a developer, not being a coder, you know, there's this other term out there about vibe coding. My experiments in vibe coding have been really, really dissatisfying. Uh, vibe coding is not there yet for someone who knows nothing about coding to be able to use it effectively and get what you want out of it.

Carolyn Woodard

It sounds like this plug-in for Claude is trying to anticipate the types of things, connections you would want to make to these other tools so that Claude can help you with those connections and with that quote-unquote coding that you might have to do, or Claude can take that on. Uh,

Carolyn Woodard

It's worth emphasizing that in their announcements about this, which I will share in the show notes, uh, every action Claude takes through these connectors requires approval.

Carolyn Woodard

So if you've been hearing about agent AI and thinking, that sounds terrible. I don't I don't want an AI agent like thinking for itself and doing stuff on its own that I'm not involved in, that I don't have the final say on, uh, then this is to address those fears and concerns, legitimate concerns, that Claude will make the connection, make a proposal of what it suggests that you should do, and then you approve it. And this is built in to this plugin.

Carolyn Woodard

So they have these pre-made workflows, and within the pre-made instructions to these agents, there is this human approval task that's built in. Um so it's not just a nice feature, it's again trying to establish uh credibility, trustworthiness, and those security safeguards that all nonprofits are looking for, all people should be looking for. Um

Carolyn Woodard

So if your organization has the Claude for Nonprofits pricing, if you have Claude and you're not using the nonprofits pricing, make sure you look into that. It's up to a 75% discount on team and enterprise plans for 501c3 organizations, um, which makes the team plan come in at $8 per user per month compared to $25 at the standard pricing.

Carolyn Woodard

When you have a paid plan, you get Claude Chat, the regular old Claude, you get Claude Cowork, and you get Claude Code.

Carolyn Woodard

The information from Claude from Anthropic about this says that they use the same platform that Microsoft uses for its verification for nonprofit pricing. If you're already verified and purchasing Microsoft nonprofit pricing licenses, then you are set to go. It should be pretty easy.

Carolyn Woodard

But if you're not using Microsoft, if you're on a different platform, if you're on Google Workspace, uh it shouldn't take very long to verify that you're at qualified 501c3.

Carolyn Woodard

And the really cool thing about this, while I'm talking about it, is that there is a free AI Fluency for Nonprofits course available through the Anthropic Academy. And this new plug-in, the Claude for Small Businesses, also has training and AI fluency also built into it. So they're really stressing the availability of these courses that help get your staff up to speed on using the AI tools that come with this bundle.

Carolyn Woodard

I believe I also saw in the press release that they're taking that show on the road at several cities around the country. They will have Anthropic Academy Claude for small business, you know, shows and um, you know, workshops and seminars, webinars, uh, that sort of thing to help small businesses understand the value of this plugin uh that they can now get. So I will share that information with you in the show notes, but really exciting. Um,

Carolyn Woodard

Not promoting one AI over the other AIs that are out there. There's lots of tools out there. I just, especially for nonprofit audiences, Anthropic is really making a lot of good moves around pricing, around training, and around the features that are available. So check it out if that is what you are interested in.

Carolyn Woodard

Second thing that was announced and that I thought was interesting for us is the ChatGPT upgrade that just happened as well. So, question for you Do you know if you're using ChatGPT what version you're using? Uh, does your staff know what version they're using? If you're just going out to the website and using ChatGPT, that's the freemium uh version, which is not secure, doesn't have a lot of the business terms and conditions that protect you. So you can purchase ChatGPT outside of Microsoft.

Carolyn Woodard

So Chat GPT's uh model is what Copilot is based on. They have this agreement, sharing agreement, speaking of open AI. And uh you can also purchase tiers outside of Copilot.

Carolyn Woodard

We've talked before about AI use at nonprofits, AI use in general is kind of going in lots of different directions. If you have 20 people at your nonprofit, they're probably using AI tools in 20 different ways. So uh you may have people who have purchased the tier uh of ChatGPT and not thought about if they've purchased it, if they're on the free tier. Uh,

Carolyn Woodard

This distinction matters both for your security and your risk mitigation, but also because there's a new Chat GPT model that was released across all tiers in early May called ChatGPT 5.5 instant. Sorry, a little lingo there. Probably not terribly important because there'll be a new one soon. But uh

Carolyn Woodard

The big thing that changed with this is memory. So if you're using ChatGPT, it can search past conversations, it can find uploaded files, it can uh look in connected accounts, for example, Gmail if you're using that. Uh so it's got a lot more functionality. It's clearly trying to catch up with some of the other tools that already have that persistent memory. Um, so it's very useful, it's worth understanding.

Carolyn Woodard

The tool has more context, it does mean that it knows more about you. So you need to ask your staff if they're on a free tier, if they're just going out to the website and using ChatGPT to ask it questions. Uh, if they're on a paid plan, if it makes sense for your nonprofit for people to be using Chat GPT because they're familiar with it, it's doing a job that they need done, but maybe get them on that paid tier uh if you can, and if you can afford it. Uh,

Carolyn Woodard

If you're having people who see that they have this capacity now, they may be connecting ChatGPT to their Gmail. They may be uploading sensitive files to it to ask you know questions, write proposals, write grant information, write a thank you note to a donor. So making sure that they understand that the default experience for ChatGPT just changed. Uh, and that's important just to know as an organization.

Carolyn Woodard

Of course, paid tiers of any AI tools generally come with stronger data protections, stronger organizational controls, stronger copyright protections, and clearer terms about how your input is being used, if it's going back to feed the model, or if it's protected within your cloud workspace. So make sure that you know that about what tools your staff are using.

Carolyn Woodard

And this applies across ChatGPT, Claude, Gemini, Copilot, other AI tools that your staff might be using or that you might be using.

Carolyn Woodard

And then I want to close on just a really quick um article that I read recently from a friend of the pod, Tim Lockie, whose organization, his company is called The Human Stack. He provides these trainings for nonprofits specifically. His whole kind of approach was that a lot of nonprofits can't afford consultants, can't afford big fancy assessments, can't really afford outside help to use their IT better if they have a database, if they have a CRM, if they just have set up Google Workspace.

Carolyn Woodard

And so his approach is to offer affordable training for someone on your staff around communications, talking with leadership, you know, running that database better, um, kind of sharing the burden of managing IT. So his idea is that you train up people that are on your staff already, and that helps you manage your IT better.

Carolyn Woodard

And of course, when AI came out, uh he was all over it. This is a really good management tool. It's a force multiplier. If you already can't afford outside consultants and you have one poor person that's running your database and is your accidental techie and has to talk to your leadership about what your budget should be, uh, AI can really help that person understand and you know just have more capacity, basically.

Carolyn Woodard

So he's been using AI for a long time. You can look on his site, I will share the link of where those trainings are. As I said, it's very designed to be very affordable.

Carolyn Woodard

And because he's been using AI for a while, uh, he is shared recently uh what he is doing personally with AI. And I'll share the link with you. You need to read this article. It's very short, but it was just so provocative to me because he built in a private Claude project a personal assistant.

Carolyn Woodard

And up until this point, we've been saying and kind of making sure people understand, you know, you're using Claude or AI tools, co-pilot Gemini at work. It's not private. You're on your work computer, you shouldn't be sharing personal things with it, uh, and you shouldn't be uploading sensitive information to it if you're working with you know students or sensitive populations, uh, you're donor list, like all of this personal information. You don't want to just share that with an AI tool because it becomes part of the information that that AI tool has.

Carolyn Woodard

And Tim's article just flipped that completely around. And I just find it, I'm still sitting with it and thinking about it because um Tim up to uploaded personal information about himself, you know, his neuro uh profile, his strengths finder results. And he set up this personal AI as a private Claude project, which he can do, as a critical mirror to give him feedback on how he's doing. And when he

Carolyn Woodard

He says in his piece that when he's anxious, when he's stuck, when he's in conflict, when he's not sure what to do, he asks this private Claude to help him see what he's missing. Uh, he says it often names the pattern that he's resisting. Uh he helps him discover what the real issue is, and he calls it a digital mirror with no judgment.

Carolyn Woodard

It has no relationship with him to protect, as it's giving him this feedback on, you know, kind of what he's how he's interacting, what he's worried about or upset about, and how the way he interacted with other people may have landed. And so giving him that kind of can be very critical or hard-to-hear feedback, I guess I would say.

Carolyn Woodard

And it's so interesting to me because, as I said, you know, we're so cautious about what goes into AI tools. We talk a lot about why governance matters. And then here's Tim saying he uploaded like really personal medical information to get an assessment of how his brain is working under stress, and that he couldn't get that quote-unquote ego-free feedback anywhere else. That, you know, it's not a relationship where you ask somebody for feedback, but then you might not take it the right way.

Carolyn Woodard

It's such a tension for me to think about these two 180-degree sides of this coin. And so I urge you to read this piece, give it some thought. Um, I haven't resolved it, as I said. I'm really, it's just so provocative to me. So I wanted to share it with you too and make you think about it. Um,

Carolyn Woodard

You don't have to go as far as Tim goes, of course, uploading medical uh records or other, you know, very personal documentation. But as I spoke spoke before on this uh podcast about what another person had called the "me" layer. So, what becomes really valuable to you in using AI is this layer of understanding that's personal to you. The AI tool that you're using, your instance of the AI tool that you're using, begins to understand the context about what you do at work, how you interact with it, how you like to interact with it, your voice, your what you're looking for from it, if you want it to be critical, if you want it to give you praise, like all of these different highly individualized, the quote unquote "me" layer that's so valuable over time at helping your AI tool give you really valuable outputs really quickly, because it's built up this understanding of you.

Carolyn Woodard

I heard someone say it at a conference that the more you treat an AI tool like a computer, the worse it works. The more you treat it like a person, you know, like an intern that you're training to work with you, the better it works. So

Carolyn Woodard

I think what Tim is pointing out is you don't, as I said, you don't have to go as far as Tim went, but you can ask your AI tool, the one you use, what am I missing? Um, you can give it context to give you more honest feedback. Is there a pattern that I'm missing? Is there something I haven't thought about? You can ask it to play devil's advocate with you. If you've got a project or program that you're working forward with, ask it to interrogate you, ask it to you know question you about. About weak points in your plan or your document or you know, whatever you're using it for.

Carolyn Woodard

So that I think is valuable to think about and to test out. You know, try having it ask you questions or um play devil's advocate with you. So I will share that article with you and of course the uh the Human Stack website as well for those trainings. But um

Carolyn Woodard

That's what I wanted to talk with you about this week. Um so happy to have you here. I will be back on next Tuesday with more nonprofit AI news.

Carolyn Woodard

I have gotten a couple questions in around funding for AI projects. So I'm kind of turning around, uh making that an episode. So stay tuned for that. I want to do a little bit more research. We know there's lots of money, kind of big grants coming out. But um, yeah, this is a great question of if you did want to do like training for all of your staff or you want to, you know, use some of this, you know, this Claude uh co-work plug-in for um QuickBooks, and how how would you go about getting funding for that or acknowledgement that you need funding to upskill everyone in your staff? So it's such a great question.

Carolyn Woodard

I'll be back here with it after I've given it a little bit more thought. Maybe I'll ask my AI to help me see what I'm missing and what I'm thinking about it. And of course, you can find me on Friday in the regular feed with our regular Friday podcast. Until then, take care.