AI for education: tutoring assistants, student support agents and admin automation built into your systems
AI for education at Goldenroz is custom development for edtech companies, universities, training providers and school systems: assistants that answer from your approved course material, agents that handle student and applicant questions, and automation of administrative work. Each is built with per-user permissions, audit logs and a person approving anything that affects a grade or a record. Projects start at $25,000 at a fixed price per phase after a free one-hour scoping call.
From $25,000, fixed price per phase
Updated 2026-11-14. By Shai Goldenberg
What usually goes wrong
- Student records are protected by FERPAIn the US, FERPA limits who may see education records. A vendor's tool may handle them only under the institution's direct control, which rules out pasting student data into a consumer chatbot.
- Children under 13 fall under COPPAProducts used by younger children need verifiable parental consent or school authorization before personal information is collected, and that shapes sign-up, logging and retention.
- A general model does not know your syllabusIt answers from the open internet, at the wrong level, and sometimes gives the solution when the assignment was to find it.
- The same questions arrive every termDeadlines, prerequisites, fees, transcripts and financial aid questions peak at application and registration time and bury a small office.
What is included
- Course assistantAnswers student questions from material you approve, cites the source lesson, and follows rules you set, such as hints before answers.
- Student and applicant support agentHandles admissions, registration and billing questions on web chat or WhatsApp from your knowledge base, and hands off to staff with the conversation attached.
- Drafting tools for teaching staffDrafts of quizzes, rubrics and written feedback for a teacher to edit and approve. Nothing reaches a student or a gradebook without that approval.
- LMS and student system connectionsIntegration through the interfaces your platform provides, such as LTI and vendor APIs, scoped and priced per system.
- Permissions, masking and audit logPer-user access so the model sees only what that student or staff member may see, PII masking where needed, and a record of every question and tool call.
Where AI for education pays off first
The safest first projects are the ones where a wrong answer is cheap and easy to catch. Answering administrative questions from published policy is one. Drafting material for a teacher to review is another. A course assistant restricted to approved content is a third, as long as it shows its source so a student can check.
The risky ones are decisions about people: grades, admissions, discipline, flags for cheating. We do not build systems that make those decisions without a person.
Education software development with privacy rules built in
Education software development that includes a language model has to answer three questions in writing: which student data reaches the model, which provider processes it under what terms, and who can see the logs. We document these with the design, so your privacy officer and counsel can review them before build starts.
The technical controls are ours to build: access rules enforced in the database, masking of identifiers before text is sent to a model, retention limits and an audit trail. Goldenroz is an engineering firm, not a law firm or an auditor. We do not certify FERPA or COPPA compliance.
Elearning software development for edtech products
For companies selling a learning product, elearning software development with AI is a product question as much as a technical one. Model cost per learner has to fit inside the subscription price, and the feature has to keep working when a provider is slow. We meter usage per customer and build a fallback chain across models.
Shai, our CTO, runs a production assistant on Claude with fallbacks and deterministic replies, and a support assistant with a knowledge base. We have no published education clients, and we say so. The build process is the one on our AI development page, and conversational products are covered under AI chatbot development.
How the work runs
Scoping call
Before the projectWe map the process, the systems involved and the cost of a wrong action. You get a fixed price for the first phase.
Prototype
About 1 to 2 weeksA working agent on one narrow task, run against real examples in a test environment. You see where it fails before you pay for production.
Production build
Roughly 4 to 8 weeksIntegrations, permissions, fallbacks, approval steps, logging and the evaluation set. The range depends on how many systems the agent connects to.
Launch with limits
At launchThe agent starts on a share of the volume, or with every action approved by a person. Limits widen as the logs show it holds.
What it costs
Every phase has a fixed price agreed before work starts. Model usage is paid to the model provider from your own account.
- Focused agent or agentic workflowFixed price per phaseFrom $25,000
- Larger AI platformsSeveral agents, many integrations or a full product$100,000 and up
- Scoping callEnds with a written scope and a price$0
- Model usageBilled to your own accountPaid to the provider
Who it is not for
- Automated decisions on grades, admissions or discipline with no human review.
- Individual teachers or single schools looking for a ready-made tool.
Who does the work
Shai leads the technology side: business software, SaaS and mobile apps, payments and escrow systems, Postgres security, AI agents built with Claude Code and MCP, and IT and infrastructure projects with senior infrastructure engineers.
Questions buyers ask
What does AI for education cost to build?
Projects start at $25,000 at a fixed price per phase, and larger platforms with several agents or many integrations run $100,000 and up. Model usage is paid to the provider from your own account. See pricing.
Is student data used to train AI models?
We build on provider accounts that you own, under the data terms you agree with that provider, and we select API terms under which your data is not used for training. Your counsel should confirm the agreement meets your obligations.
Can the assistant work inside Canvas, Moodle or Blackboard?
Where the platform supports external tools through LTI or an API, yes. We confirm what your installation allows during scoping. We do not claim past integrations with a named LMS.
Do you build AI grading or cheating detection?
We build drafting aids that a teacher reviews before anything is recorded. We do not build automated grading without review, and we do not build AI-writing detectors, which produce false accusations.
Do you work with individual schools or teachers?
Our minimum project is $25,000, which fits edtech companies, universities, training businesses and districts.
Guides on this topic
- Build a working MCP server in TypeScript with the official SDK, then add what production needs: permissions, PII masking, audit logs. Follow the steps.
- MCP does not replace REST APIs. It wraps them so AI applications can discover and call them. Compare MCP vs API and see when an MCP server is worth building.
- MCP is an open standard that connects AI applications to your tools and data. Learn how hosts, clients and MCP servers work, and what to secure first.
More for education
AI by industry
- AI agent development
- AI for accounting
- AI app and software development
- AI code review and AI coding training
- AI strategy and readiness
- Claude consulting and implementation
- AI for construction
- AI for enterprise
- AI for finance and banking
- Generative AI and LLM development
- AI for healthcare
- AI for insurance
- AI for law firms
- AI for logistics
- AI for manufacturing
- MCP and Claude integration
- AI for real estate
- AI for retail and ecommerce
- AI for small and mid-size business
- AI
Talk to the person who will do the work
One hour, no sales team. You talk to Shai Goldenberg, the CTO, about your systems or your product, and you leave with a clear next step and a price range.