Curriculum-Grounded AI for Higher Education
Faculty define what assistants know and how they respond. Students get course-aligned guidance. Institutions set data, model, and access controls.
Explore assistants across sciences, humanities, and professional studies. Review plans or discuss institutional fit.
One platform for course and campus work.
From the front office to the classroom. Choose where you want to start.
- Faculty Build a course assistant from the materials you already teach. Teaching assistants built on your course materials, on your terms.
- Students Study from your instructor's assistant or your own notes. Course-true help that builds understanding, not shortcuts.
- Institutions Evaluate campus AI with shared policies and local ownership. Academic departments and campus services such as admissions or IT.
Build deeper understanding.
Generating completed work short-circuits the learning journey. True mastery develops when students are provided with contextual guidance and the intellectual space to synthesize information, construct their own reasoning, and arrive at the destination themselves.
The instructor decides how the assistant responds.
Here a student asks outright for the answer: why a strike failed, why respiration needs oxygen, what a symbol means. This instructor chose to have the assistant point to the relevant sections of the course materials with citation chips, then ask one question that moves the student forward instead of stating the answer. That approach is the instructor's choice, not a method built into Alkimi.
Another instructor might have the assistant answer directly with citations, check a student's worked attempt and explain what was missed, or keep to this week's readings. Alkimi provides the tools; you provide the pedagogy.
How to read this example
- This is the instructor's own test of the learner view before sharing, on short demo lecture notes. Your assistant works from the materials you select.
- A citation makes a claim inspectable, not correct. Responses can be wrong or ignore an instruction, so test representative questions and revise sources or instructions before sharing.
Or walk through the full faculty setup: sources, instructions, testing, and sharing.
Answers anchored to your materials.
Generic AI gives confident answers with no connection to what is actually taught. Uploading a file is the easy part. What matters is what happens next: sources stay connected, guidance is configurable, and sharing happens inside institutional controls.
- Keep course materials connected to the assistant. Selected documents, pasted text, and website content stay connected, so responses can follow your course notation and terminology and cite the sources you chose.
- Choose how it supports learning. Configure teaching instructions, source scope, and web access, then test the resulting behavior before sharing.
- Institutional guardrails, faculty autonomy. Combine local course setup with institutional controls for models, permissions, and sharing. Hand out dedicated links or use supported website and LMS embeds.
Coursework, too
With the relevant tools enabled, draft and revise quizzes or rubrics for supported export formats.
Rich output where the subject needs it
Read equations and proofs, code, data tables, and quoted passages in context.
Built with educators
Alkimi grew from concrete classroom and campus problems. Meet the team
Know how your materials and conversations are handled.
Alkimi documents how customer content is stored, processed, shared, and deleted. Institutions remain custodians of student education records, and users control explicit sharing. Each claim below links to where it can be checked.
- No training on your content. Alkimi does not use customer content to train models, and model providers are configured for no-training use. No-training does not mean no processing or retention. Read the security overview
- Third-party FERPA attestation and SOC 2 Type 1 report. Reports are available on request through the Trust Center. See compliance details
- Permissions and explicit sharing. Chats and materials are private by default; sharing is an explicit choice within organization policy. Review sharing controls
- TLS 1.2+ in transit and AES-256 at rest. Encryption details are documented alongside retention and deletion practices. Read the security documentation
Reports and questionnaires are requested through the Trust Center(opens in a new tab) . The FERPA attestation is third-party, not a certification, and the SOC 2 report is Type 1.
Frequently asked questions
Questions about Alkimi
How Alkimi differs from a general assistant, how to get started, how students access an assistant, and what happens when a response is wrong.
How is Alkimi different from a general-purpose AI assistant?
General-purpose assistants begin with broad model knowledge and general conversational behavior. Alkimi combines selected course or campus materials with teaching instructions, safeguards, model controls, permissions, and usage reporting for higher education. Faculty configure what their assistants know and how they respond, while administrators enforce organization-wide model, access, and sharing policies.
How can our team get started?
Students and individual educators can create an account and choose a self-service plan without an institutional contract. Departments and institutions can work with the Alkimi team on a bounded pilot, security review, identity and LMS fit, onboarding, and rollout planning before expanding.
How do students access an assistant?
Students can open an assistant from a direct URL, an invitation, or an approved website or LMS embed. Depending on the institution's settings, access may require an account, use an external invitation, or allow organization-funded guest use.
What should I do if an assistant gives an incorrect answer?
Check the cited source before relying on an important response. Assistant owners can test the question, review the selected materials and retrieval settings, refine the instructions, or try another institution-approved model. Alkimi improves traceability and control, but it does not guarantee that every generated answer is correct.
Does Alkimi replace our current LMS?
No. Alkimi works alongside existing tools through dedicated assistant links, website embeds, and iframe placement where the destination allows embeds. Institutions can add guided AI assistants without moving course content or campus operations into a new LMS.
What does "curriculum-grounded" mean?
Curriculum-grounded AI answers from materials selected by an instructor or institution, such as syllabi, textbooks, lecture notes, and policy handbooks. Responses can cite the document, page, and passage they used, and owners can control how broadly an assistant searches those sources. Grounding makes answers easier to verify, but important decisions still need human review.
Can instructors control whether an assistant gives students direct answers?
Yes. Instructors can start with guided or Socratic teaching templates, then adjust the instructions, tone, safeguards, source use, and amount of help the assistant provides. They can configure an assistant to ask follow-up questions, reveal information gradually, or resist direct homework answers. Instructors should test representative prompts before sharing an assistant because generated behavior can still vary.
How is customer content handled?
Institutions remain custodians of student education records, and users control explicit sharing. Alkimi does not use customer content to train models, and model providers are configured for no-training use. Security documentation, subprocessors, retention practices, and encryption controls are public, while the third-party FERPA attestation and SOC 2 Type 1 report can be requested through the Trust Center.
Start with your own course materials.
Bring one reading, configure how the assistant should guide students, and review its responses before you share it.
Create an account, verify your email, then set up your assistant. Review faculty plans.