Every student deserves the chance to learn with great support. Every teacher deserves more of it, too.
America’s teachers are being asked to do more at a time when many schools are struggling to find enough people to do the work.
The Learning Policy Institute’s 2026 analysis estimates that at least 425,412 teaching positions across the United States were either unfilled or filled by teachers who were not fully certified for their assignments—roughly 1 in 8 teaching positions nationally. And because not every state reports complete data, the Institute cautions that even this figure is an undercount. Learning Policy Institute
The problem is not confined to one part of the country or one kind of classroom. Every state and the District of Columbia reported shortages in multiple teaching areas. Special education, science, and mathematics are among the most widespread shortage areas, while the deepest numbers of shortage positions include special education, elementary education, language arts, and career and technical education. Rural communities face particular challenges filling vacancies. Learning Policy Institute
The consequences ultimately reach students.
Schools confronting shortages may increase class sizes, rely on substitute or underprepared teachers, cancel courses, or place additional responsibilities on the educators who remain. And the burden is not distributed equally: students from lower-income backgrounds and students of color often have less access to certified and experienced teachers. Learning Policy Institute
We built Teach-1 because we believe artificial intelligence can help.
Not by replacing teachers.
By giving them more support.
More time for teaching. More support for learning.
A great teacher does far more than deliver information.
Teachers understand their students. They motivate them. They recognize when something is wrong. They build relationships, create communities, exercise judgment, and help young people become more than the sum of what they can demonstrate on a test.
Teach-1 is not intended to replace that relationship.
It is being developed to help extend what great educators can do.
Imagine a teacher introducing a difficult mathematics concept to 30 students. Some understand immediately. Some need another example. One needs to return to a concept from two years earlier. Another is ready to move ahead.
A teacher cannot simultaneously provide 30 individual lessons.
Technology can help.
Teach-1 is a specialized tutoring model designed around teaching and learning rather than general-purpose conversation. It can work with a student individually—asking questions, identifying misunderstanding, explaining a concept differently, providing additional practice, and helping the student continue learning.
The teacher remains the teacher.
But every student can have more opportunities to receive individual help.
Why an open model?
Access matters.
Some of the schools and communities that stand to benefit most from new educational technology are also those least able to afford expensive technology at scale.
That’s why Teach-1 is being developed differently.
Teach-1 is designed to deliver advanced pedagogical intelligence at a fraction of the cost and compute of much larger general-purpose models, while remaining self-hostable on modest hardware.
We believe schools, universities, researchers, educators, and communities should have meaningful choices about how educational AI is deployed.
An open, efficient model can make it possible to experiment locally, conduct independent research, adapt technology to different educational environments, better control infrastructure and data, and bring advanced tutoring capabilities to places where continually paying for frontier-scale inference may not be practical.
The goal is simple:
Make excellent educational AI available to more people, not fewer.
AI is not the solution to the teacher shortage.
Teacher shortages have deeper causes.
The Learning Policy Institute points to a shrinking pipeline into teaching and high teacher attrition. It recommends investments in strong and affordable preparation pathways, mentoring for beginning teachers, competitive compensation, improved working conditions, and better workforce data. Learning Policy Institute
Those problems require people, institutions, funding, and public policy.
A model cannot fix them.
But while America recruits and retains more great teachers, we can also build tools that help the teachers already serving our students accomplish more.
We can give a student another explanation when the teacher is helping someone else.
We can give teachers another resource for differentiated instruction.
We can make high-quality tutoring more accessible beyond families who can afford private tutors.
We can give rural and underserved schools access to capabilities that should not be reserved for the wealthiest school systems.
And we can build those capabilities openly enough that educators and researchers can help determine what educational AI should become.
Technology in service of teachers
There is a version of the AI future in which technology is primarily used to reduce the number of people involved in education.
That’s not the future we’re building Teach-1 for.
We’re interested in a different possibility:
What if AI helped a great teacher reach further?
What if every child could ask for another explanation?
What if students could practice without being embarrassed about getting something wrong?
What if a teacher could more easily support students learning at very different levels?
What if advanced tutoring were available in a rural school as readily as in one of America’s wealthiest districts?
And what if educators, schools, universities, and researchers could participate in building and improving the technology that makes that possible?
Those are the questions behind Teach-1.
America needs more great teachers. We should invest in them, support them, and make teaching a profession that talented people want to enter and remain in.
At the same time, we have an opportunity to give those teachers capabilities previous generations could never have imagined.
Teach-1 is our contribution to that effort: open pedagogical intelligence built to help teachers teach more and help more American students get the individual support they deserve.