Why AI Policy Looks Different for an AMC 

Part 2 of an ongoing series on AI strategy and leadership at AMR Management Services 

If you work at a traditional company, AI policy is mostly about what tools your team can use and what kind of information they can put into those tools. That is important, but it is relatively straightforward. 

At an association management company, the challenge is more complicated. And understanding why starts with understanding how AMCs like AMR are actually structured. 

The Shared Services Model 

AMR operates under a shared services model. That means in addition to the account team, our extended team is made up of specialists: CPAs, graphic and website designers, event managers, membership experts, etc. who work across multiple client associations. 

This structure creates real value for our clients. A small professional association can have a credentialed CPA managing their finances without paying for a full-time finance director. An executive director on our team might support two associations, bringing experience and best practices across  both of them. Clients get expertise they could not afford to hire on their own, and they only pay for the portion of that person’s time their association actually needs. 

But it also creates a responsibility that shapes everything we do, including how we approach AI. 

Accreditation Means Separation Is Not Optional 

AMR is charter accredited through the AMC Institute, which sets professional standards for how association management companies operate. One of those standards is that client data must remain separate. No commingling of bank accounts or databases. Each client association operates as its own distinct entity in our systems. 

That standard exists for good reason. Our clients trust us with their member data, their finances, their board communications, and their organizational history. A breach of that separation is not just an operational problem. It would be a violation of the trust and commitment to partnership that makes the AMC model work. 

So, when we started thinking seriously about AI, the question was never just “what tools should we use?” The question was “how do we use AI tools in a way that holds up against the same standard we apply everywhere else?” 

Projects Are a Good Start 

The good news is that the major AI platforms are moving in the right direction. Both Claude and ChatGPT now have functionality called Projects, which allow you to create separate, contained workspaces for different clients or purposes. Within a project, you can load specific context like a client’s brand voice, mission statement, communication style, and strategic priorities, so that the AI’s outputs are grounded in that client’s world. 

This matters more than it might seem at first. Consider a team member who works with two very different clients. Without separate project workspaces, an AI assistant might blend context from both, starting a draft email with “Hi Dr. Smith” for a client that serves accountants or government agencies, not physicians. That kind of error is more than a nuisance. It would signal to clients that their work is not being handled with the care and attention they are paying for. 

Projects help prevent that. They are a meaningful first step toward building client separation into how our team uses AI day to day. 

The Harder Problem 

The more complex challenge is what comes next. AI tools are increasingly able to connect to other platforms: your email, your design tools, your accounting software, and your membership database. That kind of integration is where AI starts to deliver real operational value, not just better drafts. 

But for an AMC, that integration layer is where the value gets very real and the separation question gets genuinely difficult. A team member supporting three clients might work across three different Canva accounts, three QuickBooks Online instances, three email platforms, and three association management databases. Connecting AI to any of those in a way that keeps each client’s data contained, and makes sure the AI is only drawing on the right client’s information at any given moment, is a challenge we are actively working to solve 

One area we are actively exploring is third-party connector platforms like Zapier and other MCPs, which are designed to bridge AI tools with the other software your team already works in. The appeal is that connectors can create structured, rules-based pathways between tools, which may offer a way to maintain client separation while still unlocking the kind of deep integration that makes AI genuinely useful day to day. 

What We Are Committed To 

AMR does not have a perfect answer yet. What we have is a clear standard: whatever approach we land on must hold up to the same separation requirements we apply everywhere else. We are not going to move fast on AI integration at the expense of the trust our clients place in us. 

As we work through the policy questions, pilot new tools, and figure out what responsible AI integration looks like for a shared services AMC, we will keep sharing what we learn. We suspect we are not the only AMC asking these questions, and the more we can work through this openly, the better it is for the whole industry. 

If your association or AMC is navigating the same terrain, we would like to compare notes. Follow along as we continue sharing what we learn in this series.