Updates on AI for Community
As always, we’ve got a lot going on…
- AI In Depth class — June 16th & 17th. If you’ve been through AI Essentials and want to go deeper — customizing AI for your org, running deep research, working with agents, and drafting an AI policy — this is your class. Two sessions, 9:30am to 11:30am Pacific each day. Register here.
- AI Essentials class — July 8th & 9th. Our hands-on introductory class for nonprofit staff, no technical background required. 9:30am to 11:30am Pacific each day. Register here.
- Free online meetup — July 22nd, 10am Pacific. This one’s going to be a real conversation: “Hard Conversations About AI.” Specifically, how do you talk about AI at your nonprofit without triggering the “are we replacing people?” alarm? Ken will share real framing examples, but most of the time is yours — bring a specific scenario you’re wrestling with. Sign up here.
- Meetup videos, slides, and resources are now archived. We’re adding past meetups to aiforcommunity.org/meetups — with videos, slides, and related materials for each session. Topics covered so far include creating presentations with AI, a gentle intro to vibe coding, and deep research using multiple AI services. We’ll be adding all future meetups there too, so if you ever miss a session, you can catch up on your own schedule.
- We’re doing a lot of in-person and partner training — and the schedule is filling up fast. This month Ken is presenting at the TechEquity AI Forum in Menlo Park (June 30th). In July: a presentation at BYLT’s board and staff retreat (July 10th), AI Essentials for the YMCA (July 13th), AI In Depth in Berkeley (July 14th), and AI Essentials and AI In Depth for the Central Valley Community Foundation in Fresno (July 20th and 21st). Further out we have training for the Stanislaus Community Foundation in Modesto (September 2nd), and a plenary and training session at the California Waterkeepers conference in October. If your organization — or one you’re connected to — would benefit from a customized in-person session, reach out and let’s talk.
- We’re looking for part-time admin help. Our program coordinator Emily has been amazing (thanks Emily!), but she needs to step down in the next 4-6 weeks, and we’re looking for someone to step in and help with logistics, scheduling, and communications. If you know someone who might be a good fit — ideally someone who’s passionate about nonprofits and curious about AI — please have them reach out to me at mailto:kkrugler@aiforcommunity.org.
- If you know anyone at a community foundation or large nonprofit who’d benefit from AI training, the best thing you can do is introduce them to us. A quick Zoom call is all it takes — schedule one here.
Hot Off the Press
It’s been a big few weeks. Here’s what matters most for nonprofits:
Anthropic (Claude)
The big news is Claude Fable 5, which launched earlier this month. Ethan Mollick (a Wharton professor who does some of the most useful independent AI testing out there) got early access and described it as “a very real leap over every model I have used before.” His tests showed it could work for up to twelve hours executing on a complex specification, including auto-launching many sub-agents — which gets at what agentic AI really means in practice. For everyday nonprofit use, the immediate wins are better writing, more nuanced analysis, and significantly better performance on longer, more complex tasks. Sadly, the US government just forced Anthropic to block access to this most advanced model, so it’s offline for now.
OpenAI (ChatGPT)
ChatGPT now works inside PowerPoint (in beta). If your team spends time on board presentations, grant decks, or donor reports, this is worth checking out — you can create new slides from source material, polish up existing decks, and ask questions about structure and flow, all without leaving PowerPoint.
OpenAI has also been working on improved memory for ChatGPT — a system called “ChatGPT Dreaming” that’s designed to make the AI’s memory of your past conversations fresher and more relevant over time.
Google (Gemini)
Google launched Gemini 3.5 Flash, a new model focused on longer, more complex tasks. And Gemini is getting real-time speech-to-speech translation across 70+ languages via the Live Translate feature — not directly useful for most nonprofits, but a good indicator of where the technology is heading for organizations serving multilingual communities.
The Bigger Picture
Pope Leo XIV released a document on the ethics of AI this month. It covers environmental impact, the risks of algorithmic decision-making, and how AI tends to amplify the power of those who already have it. It’s written accessibly and is worth a read regardless of your religious background — a link is easy to find by searching “Pope Leo AI encyclical.”, and I found Katie McGrady’s thoughts on it useful.
There’s also a noteworthy paper from economists at the University of Virginia and Anthropic estimating that the US “AI economy” is growing at roughly 2,600% per year in quality-adjusted terms. The catch is that this growth is largely invisible in standard economic statistics because AI capabilities are improving so fast that prices are falling as fast as quality rises. The practical implication for nonprofits: AI tools are going to keep getting better AND cheaper. The value of building AI skills now is that you’re ahead of the curve when that happens.
Tip of the Month
Here’s a habit worth building into your AI workflow, especially when you’re stuck or not getting great results: always ask the AI for feedback on your prompt before running the actual task.
The idea comes from a simple observation — we’ve all had the experience of typing a quick question into AI and getting a generic, unhelpful response. Usually that’s not because the AI can’t do better, it’s because we didn’t give it enough to work with.
So before you ask AI to draft that grant narrative, write that donor appeal, or analyze those survey results, try this first:
“Here’s my prompt: [paste your prompt]. What’s missing or unclear? What context would help you give me a better answer? What would you ask me before starting?”
The AI will often surface things you hadn’t thought of — your target audience, the tone you want, constraints you’re working within, what “good” looks like for this specific output. You can then either incorporate that feedback into a revised prompt, or simply answer the questions in your next message.
A few things to keep in mind: this works best with paid versions, since the free tiers are more likely to just tell you the prompt is fine. And it’s most valuable for higher-stakes work where you really need a good first draft, not quick one-off questions. But for those moments — writing a board report, crafting a complex volunteer brief, analyzing your program data — it’s a genuinely useful extra step that takes about 30 seconds.
Final Thoughts
I’ve been thinking about what actually makes people successful with AI. After running dozens of workshops and talking to hundreds of nonprofit staff, I’ve noticed the same five things come up over and over.
Curiosity is the foundation. People who get the most out of AI are the ones who try things, break things, and ask “what happens if I do this?” You don’t need to be a technologist. You just need to be willing to experiment.
Willingness to iterate is what separates good results from mediocre ones. Your first prompt almost never nails it, and that’s fine — that’s how AI is supposed to work. The people who get the most out of it treat it as a conversation, not a vending machine. You draft, it responds, you refine. Three or four rounds of that usually gets you somewhere genuinely useful.
Skepticism keeps you honest. AI is confident even when it’s wrong. Treating every output as a draft to be reviewed — not a finished product — is still the right mental model, even as the models keep improving.
Ability to clearly communicate matters more than any fancy prompting technique. I sometimes call AI “the revenge of the liberal arts degree”. If you can write a clear email, you can write a good prompt. The skill is the same: state what you want, for whom, in what tone, with what constraints. If you can do that in plain English, you’re most of the way there.
And finally, management skills — with current AI models, everyone is essentially a middle manager, and AIs are your team. Being able to clearly define a task (including what constitutes “done”) is becoming critical. A key part of that is knowing what information matters and making sure AI has it. The single biggest lever you have is giving AI relevant context: your organization’s mission, your audience, past examples of work you liked, constraints you’re working within. This is essentially the same skill as briefing a new staff member or contractor. The better your briefing, the better the output. AI just needs that context spelled out a little more explicitly than a human colleague would.
None of these require a technical background. They’re skills any thoughtful nonprofit professional already has in some form. AI just gives you a new place to apply them.