Chatbots, add-ons or a connected workspace: which kind of AI are you buying?
I had a great conversation recently with a tech startup founder and a nonprofit executive director. At one point the founder asked me, “Do you support tools?”
He explained how he sees AI right now. It’s either a feature added to software you already use, like the AI add-on in Monday.com, or a single-purpose app built for one job, like Lovable, which builds web apps from a plain-language description. In his view both are useful, but each one keeps your data in its own separate room.
That stuck with me. If a tech founder feels like AI is scattered across a dozen separate features, imagine how it feels to a nonprofit leader.
Most nonprofits don’t need a “summarize with AI” button in five different subscriptions, and they don’t need ten separate AI tools to manage. What they need is something that sits above all of it. I call it a harness. It connects your CRM, your accounting data, your marketing tools and your team’s knowledge into one coordinated workspace. An add-on only knows what’s in its own app, while a connected workspace can see across your organization.
Why an add-on only sees one app
When a project management tool adds an AI assistant, that assistant works inside a tightly limited view. It sees the rows and columns in that tool and nothing else.
A connected workspace works differently in three ways:
- It connects across your systems. It links to your CRM, cloud storage and communication tools through secure, approved connections.
- It can act across them. An add-on can only do things inside its own app. A connected workspace can pull donor data from your CRM, run an analysis in a sandbox and put the results into a document in your shared drive.
- It remembers. A plain chatbot forgets you the moment you close the tab. A connected workspace keeps a lasting record of your brand rules, your team’s preferences and how you do things.
The three levels of AI tools
If you’re looking at AI for your organization this year, it helps to know which level you’re actually buying.
- Level 1: the chatbot. ChatGPT, Claude or Copilot in a browser tab. It has no live connection to your data or files, so everything goes in and out by copy and paste.
- Level 2: the app add-on. AI buttons built into tools like Monday.com, Notion or Canva. These are handy for work inside that app, but they can’t see anything outside it.
- Level 3: the connected workspace. A layer that sits above your tech stack, connects several data sources, remembers your organization’s context and runs multi-step work across departments.
Most organizations are at level 1 or 2. Level 1 keeps your staff doing a lot of manual copying, and level 2 tends to speed up the silos you already have. Level 3 is where a team actually gets capacity back.
How to work out where you are and what to do next
- List every AI feature you’re paying for. Include add-ons bundled into subscriptions you already have. Note who uses each one and for what.
- Track the copying for a week. Ask staff to jot down each time they paste something into an AI tool or move AI output into another system. That list shows you where the gaps between your tools are.
- Pick one workflow that crosses systems. Event registrations that end up in your CRM and then in a thank-you email are a good example. So is pulling donor numbers into a board report.
- Ask vendors what their AI can see. Which systems does it connect to? Can it take action in them, or only read? What does it remember between sessions?
- Keep the add-ons that earn their place. A design assistant inside Canva is still useful for design work. Just don’t expect an add-on to do cross-system work it was never built for.
For leaders: know which level you’re paying for
Before you sign up for one more AI feature, find out where your team’s AI time actually goes. These are the questions I’d start with:
- Is my team spending most of its AI time in chat windows, in add-ons or in connected workflows?
- How many separate AI subscriptions are we paying for?
- Which of our most important workflows cross two or more systems?
- Would the next AI tool we’re considering connect anything, or would it add one more silo?
If the answers show a lot of copying between tools, the next step is something that connects them, rather than another add-on.
That third level is what I’m building with Cosly, an AI staff team for small nonprofits.