AI in the District.

Eighteen months ago, "AI" was a marketing word vendors used to differentiate. Today it is a procurement category with its own review path, its own veto authority, and its own vocabulary. Districts that did not have an AI policy in 2024 have one now. Districts that already had one have rewritten it twice.

For this report, we sat in on (or reviewed transcripts of) 92 district AI policy and procurement conversations between October 2025 and March 2026. The buyers were curriculum directors, CTOs, data privacy officers, superintendents, and — in nine cases — school board members. The vendors were a mix of incumbents and start-ups. The patterns were remarkably consistent.

The three questions every reviewer asks

Across all 92 conversations, the same three questions surfaced. If your sales motion does not have a crisp, written, board-ready answer to each, you are losing time and deals you do not need to lose.

1. "What student data leaves our walls, and where does it go?"

This is a data flow question, not a privacy question. Reviewers want a diagram. They want named sub-processors. They want to know whether student work is used to train models, by whom, and under what contract. "We are SOC 2 compliant" is not an answer; it is an evasion that gets noticed.

2. "What happens when the model is wrong?"

Reviewers have moved past asking if models make errors. They want to know your error surface, your human-in-the-loop design, your appeals process, and your incident history. Vendors who answer with confidence and specificity win this question. Vendors who answer with reassurance lose it.

3. "Who is accountable if a parent calls?"

This is the question that quietly kills more deals than the first two combined. Districts want to know which named human at your company a superintendent can call when a parent escalates. They want that name in the contract. Vendors who cannot supply it are read as immature, regardless of product quality.

"The vendors who won our review weren't the ones with the best AI. They were the ones who treated our policy team like adults and gave us answers we could defend in public."

— Chief Academic Officer, urban district, 64K students

What buyers actually want (and are not getting)

• →A one-page AI fact sheet they can attach to a board memo without editing.
• →A model card in plain language describing what the AI does, what it does not do, and known failure modes.
• →A standard DPA addendum for AI that pre-answers the eight questions every district counsel asks.
• →A reference list of districts that have completed the AI review and would talk to peers.
• →A standing offer to brief the cabinet directly — without the AE in the room — when the deal warrants it.

Three openings most vendors are missing

• →The AI policy itself is a wedge. Offer to review the district's policy with their team. Vendors who help districts write better policy become trusted; vendors who only react to it stay transactional.
• →Curriculum leaders are under-served. The conversation has skewed to CTOs and privacy officers. The buyers with the most actual influence over adoption — curriculum directors and instructional coaches — are starved for honest AI guidance.
• →Procurement loves a vendor with a process. A standard AI evaluation kit, sent unprompted, signals maturity and shortens cycles by an average of 23 days in our sample.

Action checklist

• →Build the one-page fact sheet this week. Have a sitting superintendent review it.
• →Name the accountable human in your contract. Put their email in the MSA.
• →Publish your model card publicly. Buyers Google for it; right now they are not finding it.
• →Train every AE on the three questions above until the answers are reflexive.

The districts in our sample are not anti-AI. They are pro-defensible. Give them what they need to defend the decision and the deal closes itself.