The number itself isn't what matters—though 107,000 students using an AI teaching assistant in a single year does tend to raise eyebrows. What's more revealing is how those students got there, and what they're actually doing when they fire up the software.
Top Hat, the Toronto-based company that's been chipping away at the classroom engagement problem since 2009, saw usage of its AI assistant Ace more than double last year. From 50,100 users in 2024 to over 107,000 in 2025. That's the kind of curve venture capitalists sketch on whiteboards, the kind that suggests something beyond novelty adoption.
But here's the thing about education technology: growth numbers can obscure whether you've built something genuinely useful or just ridden a hype cycle. Plenty of AI tools in higher ed are still searching for their actual purpose, slapping "powered by AI" on features that don't meaningfully change how teaching or learning happens. Top Hat's approach has been different—almost cautiously incremental. Launch quietly in November 2023, embed it directly into an existing platform that already had institutional buy-in, then iterate based on what professors and students actually needed.
Two years in, Ace has become something worth examining not for revolutionary promises, but for what it reveals about the messy reality of deploying AI in institutional contexts where accuracy matters and hallucinations can torpedo credibility overnight.
Two Tools, One Platform
The architecture is straightforward, almost obvious in retrospect. Professors get one set of tools, students another, both working from the same underlying content.
On the instructor side, Ace generates assessment questions, discussion prompts, and interactive elements by digesting whatever course materials a professor feeds it—lecture slides, assigned readings, homework problems. The Lecture Enhancer feature, which arrived in March 2024, lets faculty upload a PowerPoint deck and receive AI-generated questions they can embed directly into class sessions. Top Hat says fewer than 5% of those suggested questions get edited before deployment, a stat that either reflects impressive generation quality or lowered faculty standards. Probably the former, given how quickly bad AI outputs get abandoned.
The system uses what Top Hat calls "in-context learning," analyzing not just text but also the imagery within slides. Whether that actually improves question quality or just sounds good in marketing materials is harder to verify from the outside.
Students, meanwhile, get an on-demand tutor that answers questions specifically about their course materials—not generic textbook knowledge scraped from the internet. They can launch AI-generated practice sessions from assigned pages, receiving immediate feedback on mobile devices. The study assistant lives inside Top Hat's interactive eTexts and custom content pages, which means it knows what a particular professor assigned to a particular section in a particular semester.
That specificity matters. Generic AI tutors work off general knowledge bases; Ace works off the syllabus. It's a narrower approach, but potentially more defensible in markets where institutional integration and course-specific accuracy count for more than breadth.
In June 2025, Ace integrated with Aktiv Chemistry, the STEM platform Top Hat acquired back in December 2022. This version provides step-by-step tutoring and personalized remediation plans, leveraging Aktiv's computer algebra system to ensure mathematical precision. Chemistry equations don't forgive approximations, and hallucinated formulas are worse than useless—they're actively dangerous to student understanding. The algebra system acts as a guardrail.
Compliance Creates Opportunity
The most recent addition arrived in October 2025 with explicit positioning around a looming deadline that has higher education institutions scrambling. Content Enhancer lets instructors upload PDFs or Word documents, then flags accessibility issues against WCAG 2.1 AA standards, suggests fixes, generates alt text and long descriptions, and adds formative questions.
The timing is no accident. The Department of Justice's Title II ADA web accessibility rule requires compliance by April 2026 for entities serving 50,000 or more people. Many colleges and universities have accumulated years—sometimes decades—of course materials that don't meet accessibility standards. PDFs without proper tagging, images without alt text, videos without captions. The remediation work is enormous, tedious, and expensive when done manually.
Top Hat is positioning Content Enhancer as an accelerator for that compliance scramble. Whether automated accessibility checking can truly replace careful manual auditing remains an open question in the field—assistive technology experts tend toward skepticism about AI-generated alt text, for good reason. But the feature demonstrates how regulatory pressure creates product opportunities in the education infrastructure market. Compliance deadlines concentrate minds and loosen budgets.
The Numbers Behind the Curve
By September 2024, Top Hat reported that more than 40,000 educators and students had used Ace. Their year-end review offered more granular detail: 1,500 faculty members had adopted the tool, collectively generating 30,400 questions through Lecture Enhancer alone. Professor usage climbed 182% year-over-year from 2023—a meaningful jump, though from a small base.
The leap to 107,000 students in 2025 represents the kind of scaling that matters for platform economics. Top Hat charges students $35 for four months of access, $58 for one year, or $96 for four years. Some institutions negotiate enterprise agreements that eliminate student fees entirely. The University of Wisconsin-Madison signed one such deal in January 2026, after doubling its adoption rate since 2022. The University of Iowa offered students a $10-per-semester license in 2024-2025 through institutional negotiation.
Campus-level deals vary considerably, but the model follows a familiar SaaS pattern: individual access pricing paired with enterprise contracts that reduce or eliminate per-seat costs. It's the same playbook that made Slack and Zoom ubiquitous—let individual users discover the value, then convert their institutions into paying customers at scale.
Crowded Field, Narrow Positioning
Top Hat isn't operating in isolation. The education AI landscape has gotten crowded fast.
Instructure launched IgniteAI in 2025, rolling it out through Canvas with backing from AWS Bedrock and OpenAI. Khan Academy offers Khanmigo at $4 monthly for learners. Chegg announced CheggMate in 2023 and added a "Create" feature in 2025 for auto-generating practice problems from student questions. Quizlet launched Q-Chat AI tutor in 2023, though notably retired the feature by June 2025—a rare public retreat that suggests the technology didn't deliver sustainable value. Pearson has been expanding AI study tools across Pearson+ eTextbooks and MyLab & Mastering since 2023. Google added AI-generated flashcards and quizzes to NotebookLM in a 2025 mobile update. SchoolAI deepened its Canvas integration in July 2025 with course TA and tutor spaces.
Nearly every major player is adding AI features. Top Hat's differentiation—such as it is—comes from tight coupling with instructor-authored content. The in-context approach reduces hallucination risk and keeps answers aligned with what professors are actually teaching, rather than what a general language model thinks is probably correct.
It's a more constrained positioning than building a general-purpose study assistant, but perhaps more defensible in the institutional market where course materials and learning management system integration carry weight. Then again, defensibility assumes competitors can't simply replicate the approach, which seems unlikely given how quickly features propagate in this market.
What Faculty and Students Actually Think
The broader EdTech AI landscape is moving fast, but adoption patterns reveal persistent friction.
A 2024-2025 research series from Tyton Partners and Inside Higher Ed found that while AI use among students runs high, many still prefer human help when they're genuinely stuck. Faculty worry about cheating, about increased workload disguised as efficiency gains, about whether these tools actually improve learning or just make assessment harder to design. Only 36% of instructors in the study reported that digital learning has led to measurable student success—a sobering figure for an industry built on promises of transformation.
Cost and access remain barriers. Many institutions still lack institution-wide access to secure AI tools, according to Inside Higher Ed reporting from April 2025. An Ellucian, UNCF, and Hampton University study focused on HBCUs found near-universal AI exposure among students and faculty, with strong demand for formal AI coursework and tools—but noted significant barriers around cost and connectivity infrastructure.
Some systems are starting to license enterprise AI assistants directly. California State University provided access to ChatGPT Edu in February 2025, signaling normalization of institution-provisioned tools rather than students cobbling together their own solutions. Academic integrity concerns and assessment redesign pressures continue to dominate faculty conversations at conferences and in department meetings.
Top Hat has published AI guiding principles emphasizing data privacy—specifically, no sharing of personal information with third-party generative AI platforms without consent—along with commitments to transparency, bias mitigation, and alignment with the NIST AI Risk Management Framework. The company holds ISO/IEC 27001:2022 certification covering its SaaS platforms and maintains encryption at rest and in transit.
Notably, Top Hat doesn't disclose which underlying LLM providers and models power Ace in its public materials. That's fairly standard practice, but it makes independent evaluation of the system's capabilities difficult.
Follow the Money
Top Hat has raised approximately $225-270 million across six funding rounds. The most recent disclosed rounds: a $55 million Series D in February 2020, a $130 million Series E in February 2021 at roughly a $500 million valuation, and a $24.5 million Series F in December 2022. Georgian led multiple rounds, betting repeatedly on the company's ability to scale in higher education—a notoriously difficult market for software companies to penetrate.
The company named Maggie Leen CEO in March 2024, promoting her from CMO. Joe Rohrlich, the previous CEO, moved to Recurly in January 2024. Top Hat acquired Aktiv Learning in December 2022 and purchased OpenClass intellectual property in July 2025 to expand what it calls "authentic, AI-powered assessment capabilities." Time magazine included the company in its 2024 "World's Top EdTech Companies" list, for whatever that recognition is worth.
Ace represents Top Hat's bet on what the company calls "human-centered AI" that scales evidence-based teaching practices—frequent low-stakes assessment, active learning techniques—rather than attempting to replace instructors outright. The positioning is careful, almost defensive. The rollout has been incremental rather than splashy. And the usage numbers suggest the product has found genuine traction, not just pilot-program curiosity.
The Bigger Question
Whether that traction translates to sustainable competitive advantage in a market where every player is frantically adding AI features remains uncertain. Network effects seem limited—there's no obvious reason a student using Top Hat's AI assistant makes it more valuable for the next student, the way social platforms compound. Switching costs exist but aren't prohibitive if a competitor offers meaningfully better features through Canvas or Blackboard integration.
The durability of Top Hat's position likely depends on execution: how quickly it can add capabilities, how well it integrates with institutional systems, how effectively it trains faculty to use the tools in ways that genuinely improve learning outcomes rather than just automating busy work.
For now, though, 107,000 students are using Ace to study. They're getting AI-generated practice problems aligned with their actual coursework, not generic materials. Their professors are embedding AI-generated questions into lectures without spending hours writing them manually. The technology is being used, not just demoed.
In education technology, that's rarer than you might think.
