This post is part of the Reimagining Learning for the Age of AI compendium from the Center on Reinventing Public Education.
When it comes to AI use in the classroom, we don’t need more tools. We need better ones. There’s plenty of innovation in ed tech, but not enough focus on quality, coherence, and the students who are furthest from opportunity.
As an early-stage investor, I see both the promise and the risk of this moment. Despite the immense number of tools flooding the market, few of them align with evidence on what works for learners. Even fewer are designed for the realities of schools serving low-income communities. This is clearly a pivotal moment, and smart investment can help turn the tide.
The Importance of Evidence
In 2024, K-12 schools spent roughly $30 billion on ed tech. By 2033, that dollar amount is expected to double. Yet only one in five of the top 150 most-used classroom tools have rigorous evidence aligned with ESSA Tier I-III standards that they positively impact students.
That reality is colliding with an urgency around learning acceleration. A recent Economist piece points to declining test scores alongside the rise of classroom computer use, suggesting technology may be part of the problem. More nuanced analyses, like Nik Bear Brown’s response (which is worth reading in full), clarify what the research actually shows and what educators know from direct experience: the impact of ed tech varies dramatically depending on the skills targeted and the quality of implementation.
“There’s plenty of innovation in ed tech, but not enough focus on quality, coherence, and the students who are furthest from opportunity.”
For example, a 2024 meta-analysis found positive effects for digital learning tools across every literacy domain, in some cases with especially strong benefits for students from low-income backgrounds. At the same time, we know that higher-poverty schools are more likely to use technology too frequently and for low-level tasks, contributing to weaker academic outcomes.
Ed tech has not failed, but its designers and implementers may have. As Brown argues, we are deploying the “wrong technology, in the wrong amounts, in the wrong ways.” Meanwhile, approaches with strong evidence remain stuck at 20% market share.
Moving Beyond the Hype
If we are serious about learning acceleration, particularly in foundational reading and math, the central challenge is aligning the promise of AI-based innovation with the needs of diverse student communities and the school systems that serve them.
Supplemental tools, such as formative assessment platforms, tutoring supports, and instructional planning aids, are often treated as add-ons rather than central levers for improvement. In practice, however, these tools can be powerful entry points for evidence-based innovation in classrooms. When well designed and aligned to high-quality instructional materials, they help educators translate strong curricula into daily practice.
For example, they can provide real-time insight into student understanding, generate targeted practice, scaffold differentiation for multilingual learners or students with disabilities, and free up teacher time for feedback and relationship-building.
Compared with large-scale curriculum overhauls, supplemental tools are relatively low-cost to implement and pilot, and under the right conditions, they can positively impact student growth.
But there is a 5% problem. Implementation makes or breaks a solution’s potential for impact. Too often, promising solutions falter not because the idea is flawed, but because schools lack the time, training, or leadership focus to use them well.
That said, there are early bright spots. In some districts, specific tools have been tightly matched to clearly defined needs, such as improving early literacy or strengthening Algebra readiness. In those cases, leaders made expectations explicit, invested in training and coaching, and monitored use over time. Teachers understood what success looked like and had practical support to get there.
“Implementation makes or breaks a solution’s potential for impact.”
Some of these tools require as little as 30 minutes per week of structured use and have been associated with measurable gains in reading and math growth. These results are most likely when expectations are clear from the start, schools receive ongoing support, and vendors share responsibility for outcomes. Approaches such as outcomes-based contracting, where payment is tied in part to agreed-upon results, can help create shared accountability and keep the focus on student growth.
When supplemental solutions are introduced without clear alignment to district goals and classroom practice, they can fragment attention and dilute impact. When they are tightly integrated within a larger instructional strategy, they can become not just “nice to have,” but reliable ways to make a measurable, scalable impact.
The Role of Early-Stage Investment
This is where I see our lane as early-stage investors. We sit close to a wide range of emerging ideas — before they are proven and before they reach scale. That vantage point gives us both opportunity and responsibility.
- Shining a light on high-potential, evidence-aligned approaches.
- Supporting context-specific research that improves products and validates potential impact.
- Encouraging co-design from day one — particularly in schools serving low-income communities.
The goal is not to invent entirely new theories of change. In foundational reading and math, we already know a lot about what works. The challenge is realizing this potential. Experience shows that early-stage innovators who are close to communities and school systems are well-positioned to identify and go deep on solving specific instructional challenges. These targeted solutions are more effective than those that are “one-size-fits-all.” Traditional investors also overlook them, so early-stage risk capital is necessary to support these new ideas as innovators build their evidence base and find product-market fit.
If we succeed as investors focused on impact and equity, we should see more high-quality solutions enter the market, out-competing (or in some cases, partnering with) existing products and maintaining results as they scale. Speed and flashy marketing may win in the short term, but quality and results will win in the long term. To meet this moment, school systems must let go of low-quality student distractions in favor of evidence-aligned solutions that truly advance reading and math growth, particularly for students who have been furthest from opportunity.
Explore the full Reimagining Learning for the Age of AI compendium from the Center on Reinventing Public Education.
