The numbers behind the mess
A district with access to 3,001 digital tools and meaningful active usage of only four is a precise description of a market that has failed its buyers on basic usability and evidence, before AI-specific efficacy questions even enter the picture. That statistic, reported by Education Week in August, sits alongside market projections showing the broader edtech sector nearly doubling from 48 billion dollars in 2024 to 90 billion by 2030, with the AI slice growing roughly six-fold over a shorter window than that.
Those two facts together describe a gold rush colliding directly with a due-diligence vacuum that nobody in the sales chain has an incentive to fill on their own. Money is flowing into AI education products faster than districts, or frankly most enterprise buyers in any industry, can build the evaluation capacity required to tell a genuinely good product from a mediocre one before signing a multi-year contract that is expensive to unwind.
Why the asymmetry persists
American Enterprise Institute researcher Mark Schneider frames the core problem as structural rather than incidental: there is always an asymmetry between what a provider knows about its own product's real limitations and what a district knows going into a purchase decision. Vendors control the efficacy data, the pilot design, and often the case studies used to sell the next renewal, while districts are left evaluating sweeping claims with a fraction of the technical staff any serious vendor employs internally.
Common Sense Media's Sue Thotz describes the practical consequence directly: the full burden of vetting AI vendors currently falls on districts that are already stretched thin across every other operational priority they carry. This reflects a market structure problem more than a judgment on any single vendor's intentions, and it explains why so many districts end up relying on vendor-supplied evidence largely because building independent evaluation capacity in-house remains prohibitively expensive for most of them.
When the asymmetry breaks badly
The AllHere case is the clearest cautionary tale the sector has produced so far. Los Angeles Unified paid roughly 3 million dollars for an AI chatbot the company ultimately could not sustain, and AllHere filed for bankruptcy shortly after the contract closed, leaving the district holding a failed product alongside a very public embarrassment. Student-monitoring vendor Gaggle is now facing separate lawsuits in Arizona and Kansas, a different failure mode entirely, but one built on the same underlying pattern of a district trusting vendor claims it had no independent mechanism to verify beforehand.
Neither case is isolated enough to dismiss as an outlier limited to one poorly managed district. Once two large, comparatively well-resourced districts have been burned by AI vendor claims within roughly a year of each other, the reasonable response from every other district's procurement office is to slow down and demand independent proof, which appears to be exactly what is starting to happen across the sector right now as renewal conversations get longer and more adversarial.
The states building a better model
Pennsylvania, New York, and Chicago have started building procurement frameworks that centralize vendor vetting rather than leaving each individual district to evaluate the same vendor claims independently and from scratch every time. That approach accomplishes two things simultaneously: it reduces duplicated diligence work spread across hundreds of separate districts, and it gives vendors a single, more sophisticated evaluator to satisfy instead of thousands of under-resourced technology directors working alone with limited time.
For vendors, a centralized framework is not automatically good news, and plenty will treat it as a threat rather than an opportunity. Closing a deal based on a favorable pilot result is harder with a state procurement office staffed by people who evaluate education technology full time than it is with a single overworked district technology director juggling a dozen other responsibilities. Expect vendors carrying weaker evidence bases to actively lobby against exactly this kind of centralization going forward.
What this means beyond K-12
The identical asymmetry problem exists in corporate learning and development procurement, just with far less public scrutiny because there is no local school board meeting where a bad AI training vendor decision gets aired in front of parents and reporters. Enterprise learning and development teams evaluating AI coaching, upskilling, or compliance training tools face the same information gap district technology directors face, they simply have fewer public examples like AllHere available to point to internally when a vendor pitch sounds too good to verify quickly, which if anything makes the discipline of independent verification more important rather than less.
The practical fix looks the same in both markets regardless of sector. Demand efficacy evidence that was not generated or funded solely by the vendor pitching the product, request references from customers who churned in addition to ones who renewed, and build procurement processes that centralize evaluation expertise rather than asking every individual buyer across the organization to become an AI evaluation specialist from a standing start with no institutional support behind them.
What to watch next
Watch whether additional states follow Pennsylvania, New York, and Chicago into centralized AI procurement frameworks over the next two quarters, since that structural shift is the change most likely to actually close the information gap rather than simply documenting its existence in another report. A wave of state-level frameworks arriving in quick succession would meaningfully raise the evidence bar for every vendor selling into education across the country, not just the ones currently under scrutiny.
Also watch for additional AllHere-style failures surfacing in smaller districts that lack the press attention Los Angeles commands. The 3-million-dollar Los Angeles contract was large enough to make national headlines, but smaller versions of the same failure pattern are almost certainly playing out in districts too small to attract coverage, and each one strengthens the case for procurement reform that vendors with genuinely strong products should welcome rather than quietly resist.
