Most executives treat standards work as background: someone else's job, finished product, read-when-relevant. That logic breaks when the standard is still being written and the agency writing it has specifically asked organizations like yours to comment before it is finished.
Two federal AI drafts are in that window right now. Both carry deadlines in the weeks ahead. Neither will wait for the next planning cycle.
What the Drafts Cover
The Commerce Department's National Institute of Standards and Technology released two AI documents this summer. Both are open for public written input. Neither is binding regulation. [2] [3]
The first, released July 29, provides guidance and templates for public-facing AI documentation, including model and data characteristics. NIST calls this a zero draft: an early standards proposal built with stakeholder input, designed to become a voluntary consensus standard if standards organizations take it up. Zero drafts are preliminary and will likely change. The comment window, a formal period for written public input, closes September 16. [2] [3]
The second, announced August 7, describes a four-stage method for testing whether an AI system works as intended and contains harm. This framework spans machine learning, large language models, multimodal systems, and agentic AI, and is designed to be flexible enough for customized assessments across all of them. NIST describes the process as systematic work to confirm that AI systems meet organizational goals while limiting negative impacts. The framework applies to organizations that conduct such evaluations or rely on the resulting reports to make decisions. Written input closes October 6. [1]
NIST frames both as voluntary. Neither creates a compliance obligation today. The proposals could affect how evaluation and documentation are discussed if standards organizations take them up, but the drafts will likely change during that process. NIST explicitly invites business decision-makers and procurement specialists who use evaluation reports to weigh in. That is a direct ask of your function. [1] [2]
The questions are practical, not ceremonial. NIST asks whether its evaluation framework is flexible enough for different systems and business contexts, which activities it misses, and what needs revision or expansion. On documentation, it asks what public information about models and data should or could be included. The agency also accepts input on future topics and how to scope them. A buyer who has struggled to explain the difference between two vendors in plain language has experience directly relevant to those questions. [1] [2]
Why Buyer Experience Is Worth Submitting
You already hold records that NIST does not: what questions you asked vendors before signing contracts, which evaluation claims you could verify and which you could not, where documentation gaps forced you to take risk on faith. NIST is constructing guidance in exactly that territory, and the agency has said plainly that buyer and procurement experience is part of what it is asking for. [1]
Zero drafts are designed to absorb practical input before formal standards development begins. That is the window: not after ratification, not after adoption, but now, while the draft is explicit that it wants operating experience. [2]
Submitting does not guarantee adoption, compliance treatment, or any downstream benefit. NIST will weigh submissions from a wide range of organizations. Buyer operating experience is one input among many.
The practical constraint is real: comments are public record, subject to public disclosure, and must exclude proprietary information. That restriction is manageable. Many procurement-level observations can be framed as professional practice rather than company secrets: what documentation categories mattered in a purchase decision; which testing evidence was legible to a business reviewer; where an evaluation report answered the wrong question. [1] [2]
Silence has a consequence too. If the resulting guidance fits vendor workflows better than buyer review workflows, buyers may face adjustment later, when the proposals are further along. That outcome is not certain. It is the avoidable risk created when the people who use evaluation reports leave the public record to the people who produce them.
What to Delegate This Week
Name one person.
That person has two tasks, both bounded in scope.
First, read both drafts against your current AI procurement and evaluation records. Not a formal gap analysis. A working comparison that surfaces where your existing process aligns with the drafts, where it diverges, and where the drafts describe something your organization does not do at all. The comparison will be useful regardless of what you submit: it tells you where your practices stand relative to the direction these voluntary proposals are moving, while any adjustment is still cheap.
Second, produce a concise, nonproprietary comment for each draft. Ground the observations in actual buying experience: what an evaluation report enabled in a purchase decision, what it failed to answer, and what documentation gap created downstream cost. Those examples make operating experience visible without promising that NIST will adopt the recommendation.
The September 16 deadline for the documentation draft and the October 6 deadline for the evaluation framework define the available window. [3] [1] If you wait until the discussion reaches an executive agenda, you may miss one window or both.
One owner. Two reads. Two short submissions. The scope does not expand beyond that unless your organization chooses to.