10 Things I Learned in 2 Years
Happy birthday, Adartova Consulting LLC! I figured it was time to synthesize some of the most important insights from my many client engagements as a way to celebrate two years of helping community benefit organizations become more efficient, effective, and happy through artificial intelligence (AI).
Yes, happy!
When used correctly, AI helps to eliminate mindless, boring, tedious, repetitive and otherwise unwanted tasks. So you can focus on why you started working for a nonprofit in the first place. Making you happier.
Top Line: I still believe more than ever than successful AI implementation is 80% culture change and only 20% technology.
The most effective organizations think deeply about transformation rather than treating AI as just another tool. Yes, it is a tool, but the changes made possible by AI can remake the day-to-day operations of your nonprofit if you let them.
Unfortunately, this transformation makes for hard work.
Fortunately, you can see some immediate benefits to this work and many long-term benefits if you persist. In the short term, my clients experience renewed energy, assistance with basic tasks, and the psychic benefits of G.R.O.S.S. (getting rid of stupid stuff). In the long term, my few clients who focused on transformation increased revenue, energized their boards and donors, made their own jobs more interesting, and began transforming the lives of the people they serve.
Making this leap doesn’t have to be rocket science. Here are some tips I have learned in the past 2 years:
Make time to fix problems. The single biggest barrier to AI adoption seems to be the constant time crunch of pretty much every nonprofit in the U.S. Why spend 2 hours automating a weekly task that only takes you 10 minutes? Because it’s an investment in future productivity that pays off after just 12 weeks.
Corollary to #1: Leadership empowers staff to invest time to fix problems. While you probably have one gung-ho AI enthusiast who supercharges her own productivity, most of your team won’t invest time to make their own lives easier unless leadership makes it a priority. That includes leading by example.
Security will only get worse. If you weren’t paying attention, in late June 2026 someone bypassed the National Security Agency’s systems in a few hours using the latest AI model. That means your own security better get locked down. Password rules, two-factor authentication, training and protection around financial transfers, and other security measures are no longer suggested. They’re necessary.
Leaders rack up incremental AI gains and track them. Most AI improvements will prove minor, but hundreds of these improvements amount to significant change. Leadership can encourage these changes by publicizing them and even by counting the number of them. I saw a nonprofit that has a fishbowl to which someone adds a marble every time they make an AI automation. It’s a visible reminder of progress that inspires more progress.
Choose just a few AI tools. Increasingly, concentrating on only one LLM chatbot (usually Gemini if you’re Google Workspace or Copilot if you’re a Microsoft shop) can accomplish most of what you need. Leaders limit the remaining AI tools so that each job does not require separate, specialized knowledge.
Cross-department collaboration kills improvements. See my “top line” above… Culture change can challenge the best organizations, and projects that require coordination across multiple departments or lines of business will take way more time. On the bright side, looking carefully at these processes can often identify waste or gaps that you should have identified years ago before we even had AI.
Specialized skills will lose relevance over time. I don’t know how to code, but asking my AI chatbot the right way enables me to code. I don’t always know what button to press on a screen, but putting a screenshot into my AI chatbot can often give me the answer. In short, all of those specialized skills that prevented nonprofits from doing things like automation, website creation, or app development can now be solved using AI. Usually for pretty cheap.
Your team is probably using unauthorized AI. Almost every business seems to experience the so-called “shadow AI” problem in which employees use AI tools they are not authorized to use. This usage causes legal risks, reputational risks, and security risks. Proper training on why this is a terrible idea can help, but you also need to offer your employees an alternative. That means a few decent, paid AI tools are no longer optional.
Everybody thinks they are behind the learning curve. AI moves way too fast to keep up. Fortunately, you don’t need to keep up! Start with the business problem and figure out the best solution, which may or may not involve AI. Don’t start with AI and try to figure out where to stuff it.
AI can be useful. This may seem obvious if you’re a believer, but I still see so many nonprofit executives who remain skeptical or even outright hostile to AI. I can now tell you hundreds of stories of major and minor successes with AI governance, training, and implementation. And my conclusion is that you should be considering AI now. It’s no longer optional.