The traditional Association Management System selection process has not changed much in twenty years. You assemble a committee, develop requirements, issue an RFP, evaluate responses, schedule demos, check references, negotiate pricing, and select a vendor. The process is thorough, defensible, and well-understood by everyone involved. It is also increasingly inadequate for the environment associations are operating in today.
The problem is not that the traditional process is wrong. The problem is that it was designed to evaluate static software, systems whose capabilities were defined at the time of purchase and changed slowly through scheduled release cycles. In that world, the RFP was a reasonable proxy for fit. If the vendor could demonstrate that their system did the things on your requirements list, you had reasonable confidence that it would continue to do those things for the duration of your contract.
AI changes this in two important ways.
First, the capabilities of AI-enabled AMS platforms are changing fast, faster than any RFP process can track. A platform that does not have a particular AI feature today may have it in six months. A platform that has it today may have implemented it in a way that creates data governance risks you have not evaluated. The point-in-time snapshot that an RFP produces is less informative than it used to be, because the thing you are evaluating is moving.
Second, the questions that matter most for AI-enabled platforms are not the questions that traditional RFPs ask. Traditional RFPs ask about features: does the system do X? AI evaluation requires asking about architecture, data practices, and governance: how does the system use member data to train or improve its models? What data leaves your environment and under what circumstances? How are AI-generated recommendations generated, and can you audit them? What happens to your data if you terminate the contract?
These are not questions that most selection committees are equipped to ask, because they require a different kind of expertise than traditional AMS evaluation. They require someone who understands how AI systems work at an architectural level, not just what they do, but how they do it and what the implications are for your organization's data, your members' privacy, and your ability to govern the system responsibly.
I have led more than 500 AMS implementations across my career, and the ones that have gone poorly in recent years have almost always involved a mismatch between what the organization thought they were buying and what the AI layer of the platform actually does. Not because vendors are deceptive, most are not, but because the evaluation process did not ask the right questions, and the vendor had no incentive to volunteer information that was not requested.
What does a better evaluation process look like? A few things.
Add data governance questions to your RFP. Ask specifically how member data is used in AI model training, what data leaves your environment, and what controls you have over that. Ask for the vendor's AI governance policy and evaluate it with the same rigor you apply to their security policy.
Evaluate the AI roadmap, not just the current feature set. Ask vendors where their AI capabilities are heading and how they make decisions about what to build. The answer will tell you something important about whether their priorities align with yours.
Bring in technical expertise for the AI evaluation. If your selection committee does not include someone who can evaluate AI architecture and data practices, bring in an advisor who can. The cost of that expertise is trivial compared to the cost of a bad selection.
And finally, weight adaptability more heavily than feature completeness. In a fast-moving environment, the ability to evolve is more valuable than a long feature list. An AMS that does 80 percent of what you need today but has a strong AI development trajectory and a governance framework you trust is a better choice than one that checks every box today but is poorly positioned for where the technology is going.
The associations that navigate this well will have a significant advantage, not just in operational efficiency, but in their ability to serve members in ways that were not possible before. The ones that apply yesterday's evaluation criteria to today's platforms will end up with systems that underdeliver and governance problems they did not anticipate.