Anthropic has expanded its Claude for Teachers platform with tools designed to help educators plan lessons and check student mathematical comprehension. Developed in partnership with Learning Commons, these updates move the assistant beyond basic chat prompts into structured instructional workflows. As school districts manage classroom AI, these features aim to combine technology with learning science.
What Happened
The two new "Agent Skills" are Lesson Preparation and Check for Understanding. Instead of requiring teachers to write complex prompts from scratch, these tools offer pre-built frameworks. Lesson Preparation is a conversational tool that helps teachers work through an existing lesson plan to anticipate student bottlenecks and plan pedagogical responses. The Check for Understanding tool generates short, targeted mathematical checks (usually one to three items) aligned with specific state standards.
This release coincides with Anthropic's expansion of Claude for Teachers as a free enterprise offering for U.S. school districts. This option allows school leaders to establish a centralized Data Processing Agreement that covers all educators in the organization. Under Anthropic's support terms, the platform does not use inputs or outputs from Claude for Teachers to train its models. It is an educator-only tool, meaning students do not log in and the platform remains compliant with the Family Educational Rights and Privacy Act (FERPA). As we previously reported, this district-controlled setup prevents educators from compromising student privacy on personal accounts.
The Bigger Picture
While classroom AI tools promise to save teachers time, researchers warn that general large language models often struggle with "curriculum cognition." According to a 2026 pre-print study, popular models frequently fail to understand prerequisite relationships and how math concepts build sequentially. To address this, the new tools draw from the Learning Commons digital infrastructure, which transforms curriculum frameworks and academic standards into machine-readable datasets to help the AI understand pedagogical sequencing.
Teachers must still verify any content these tools create. A study published in the Proceedings of the ACM Conference on Learning @ Scale analyzed over 5,600 middle-school math questions generated by AI. The researchers found that while 68.74% of the questions were ready to use, 19.42% required revisions, and 11.84% had to be discarded entirely due to errors or confusing language.
A systematic review in Discover Education noted that while traditional, adaptive digital tutoring programs show strong evidence of boosting math achievement, the empirical evidence for generative AI systems directly improving student test scores remains limited.
What This Means for Families
For parents, these updates position AI as a tool for teachers rather than a replacement. Because Claude for Teachers is not a student-facing application, children will not interact with the AI directly.
For educators, the tools streamline prep work but do not eliminate the need for human oversight. Teachers must still review the mathematical logic of every generated question to ensure assessments are fair and accurate.
What You Can Do
To manage this transition, parents and educators can take several steps. First, check with school administrators to see if they have signed up for the free enterprise tier of Claude for Teachers to ensure student data privacy is managed through a formal district agreement. Second, when using the Check for Understanding tool, apply a Generate-Filter-Edit approach by checking for overly wordy phrasing or mathematical errors before sharing quizzes with students. Finally, use the Lesson Preparation conversational skill to talk through upcoming curriculum topics and pinpoint potential student sticking points before teaching.