Agent of Change
Move over, 007.
CEOs, researchers, and tech journalists alike predict 2025 will be the year of AI agents. From what we're seeing in the AI-powered nonprofit space, it's hard to disagree. What is agentic AI, exactly? Think of it like AI graduating from taking orders to taking initiative. Rather than just responding to requests, agentic AI can independently pursue goals and make decisions.
To better understand this shift, I caught up with Jared Chung, founder of CareerVillage.org and author of The Agentic Era newsletter. As someone leading a nonprofit focused on the future of work, Jared has a unique perspective on where this technology might take us.
Out: Traditional AI Chatbots, In: AI Agents
Last week, OpenAI announced Operator, meaning all three of the major frontier AI model companies have released general purpose research demos of AI agents. Why is agentic AI suddenly blowing up? As Jared explains, “LLMs were the wick, and the first pile of dynamite is agents.” It's not that agentic systems are new; we've had them for years. What's changed is their recent increase in capability, reliability, and autonomy. Today's agents are leveling up in three key ways: they're tackling complex goals instead of just following scripts, executing multi-step tasks with less hand-holding, and managing a broader range of work. “These capabilities are undoubtedly being accelerated by the maturity of LLMs, and the results of huge investments into custom models purpose-built for agentic workflows,” Jared said in The Agentic Era.
“Because AI has the capacity to be agentic itself…you can leverage AI in two distinct ways. In some instances, you might want to work closely with an AI — such as when you’re learning a new language or practicing mindfulness skills. In others, such as optimizing your home’s energy consumption based on real-time energy prices, you might prefer to let an AI handle that by itself. Either way, the AI is increasing your agency, because it’s helping you take actions designed to lead to outcomes you desire.” - Reid Hoffman, Superagency, which was released today.
Who's Building?
The usual tech giants are leading the charge, each with their own flavor of AI agent. But Jared notes that the AI agents that are expected to change the fabric of our society don’t exactly exist yet. He test-drove Anthropic's general purpose AI agent and aptly called it "a glimpse into what might be possible." The agent completed his tasks, albeit with the pace and precision of an intern's first day: slow, supervised, and with a few creative interpretations along the way.
While most of us are waiting to see when these high-quality agents will become widely available, nonprofits aren't just watching from the wings. Jared's team at CareerVillage.org is taking a proactive approach: building services that don't scale. This might sound like startup heresy, but Jared's playing the long game. CareerVillage.org is developing high-touch career coaching programs that are too labor-intensive to offer widely right now — sort of like a chef perfecting recipes before writing a cookbook. When AI agents mature enough to replicate processes, they'll already have the winning playbook ready to roll out at scale.
At the same time, nonprofits are expected to do more with less. But with properly leveraged AI agents, “a 20-person nonprofit might be able to do what it used to take a 2000-person nonprofit to do,” Jared noted. Suzanne DiBianca, EVP and chief impact officer at Salesforce, agrees. In Forbes, she highlighted how nonprofits are already scaling impact with AI agents. When your mission is fighting climate change or providing disaster relief, that kind of multiplication matters.
Field Operatives
Take, for example, Digital Green. Since 2008, Digital Green has been helping farmers cultivate success from the ground up, and now agentic AI is taking their mission to new heights. In 2023, they launched Farmer.Chat, an AI-powered chatbot that quickly proved its value to farmers, even in its early pilot stage. To deepen impact, they went further to find real-time solutions to dynamic, unique farmer challenges. When weather patterns shift suddenly, new pests emerge, or diseases strike, farmers need actionable advice that's current and locally relevant. That's where agentic AI comes in.
Unlike traditional AI that relies on static information, agentic AI can actively pull data from multiple live sources — like current weather data, local market prices, and recent pest reports. When a farmer asks why their crops are struggling, Farmer.Chat can connect the dots: through a conversation with the farmer, it might notice they're using fertilizer that isn’t ideal for current soil conditions, recommend a better option, and even show them where to buy it locally at the best price. What might have taken IRL extension agents hours of research and travel is done in seconds by the AI agent. While the team continues to iterate and improve based on farmer feedback and rapid AI advancements, they're already growing a future where tech helps farmers work more efficiently and effectively.
New Year, New Agency
As we step into 2025, the rise of agentic AI presents both an opportunity and a responsibility for the social sector. "The first step," as Jared emphasizes, "is to make sure that the social impact sector is an informed and active voice in shaping how this tech develops.” Nonprofits aren't just potential users of AI agents — they're representatives of their communities and causes.
The key isn't waiting for perfect AI agents to arrive. It's preparing now so we can harness their potential for maximum social impact. Organizations like CareerVillage.org and Digital Green are showing us how to approach agents thoughtfully: start building the frameworks today that agents can power tomorrow, keep our beneficiaries' needs at the center, and imagine bigger than our current constraints. As Jared puts it, we need to "keep our eyes open" and track how agentic AI evolves, especially in the sectors that matter most to our missions. The future of AI isn't just about tech becoming more capable — it's about ensuring those capabilities serve everyone. To stay on top of the fast-moving agentic AI discussion, subscribe to Jared’s The Agentic Era newsletter.
Quick Bytes
Other Sector Stories
Is agency playing double agent? In Reid Hoffman's new book Superagency, he paints an optimistic picture of how AI can increase our own agency. One idea that’s stuck with me is the following equation: intelligence + energy = human progress. By augmenting both forces, AI could amplify what humanity can achieve.
Mozilla's new Llamafile turns complex LLMs into a single file that runs directly on your computer, no fancy setup needed. Think of it as AI “to-go” — always available, even offline. Perfect for organizations in low-connectivity areas or those handling sensitive data, Llamafile unlocks AI’s potential without compromising privacy or accessibility.
There is real (founded) fear that students will bond with AIs in ways that will impact their social development. The 74 highlights ed-tech leaders like Jared and Tiffany Green of Uprooted Academy who are building AI that fosters authentic human-to-human connection, not social isolation.
Lost in translation? Not if Atif Javed can help it. After watching his family and refugees struggle with language barriers, this MIT grad created Tarjimly, an AI-powered platform that connects volunteer translators with refugees in real-time. Read his story here.
Nonprofits are no strangers to the constant juggle of maximizing resources and creating impact. In this thoughtful piece, my co-founder Shannon Farley shares five actionable ways AI can streamline nonprofit operations — from data analysis to donor engagement — without losing the critical human touch.
APN Funding News
Google.org opened applications for the second cohort of the Generative AI Accelerator.
The Patrick J. McGovern Foundation announced $73.5M toward AI for Humanity.
Renaissance Philanthropy launched the AI for Math Fund, a new grant program supporting projects that apply AI and machine learning to mathematics.
Let’s Talk
I am living and breathing AI for humanity these days. If you are too, let’s talk!