GoldenrozBook a call

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 FERPA
    In 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 COPPA
    Products 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 syllabus
    It 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 term
    Deadlines, prerequisites, fees, transcripts and financial aid questions peak at application and registration time and bury a small office.

What is included

  • Course assistant
    Answers student questions from material you approve, cites the source lesson, and follows rules you set, such as hints before answers.
  • Student and applicant support agent
    Handles 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 staff
    Drafts 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 connections
    Integration through the interfaces your platform provides, such as LTI and vendor APIs, scoped and priced per system.
  • Permissions, masking and audit log
    Per-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

  1. Scoping call

    Before the project

    We map the process, the systems involved and the cost of a wrong action. You get a fixed price for the first phase.

  2. Prototype

    About 1 to 2 weeks

    A working agent on one narrow task, run against real examples in a test environment. You see where it fails before you pay for production.

  3. Production build

    Roughly 4 to 8 weeks

    Integrations, permissions, fallbacks, approval steps, logging and the evaluation set. The range depends on how many systems the agent connects to.

  4. Launch with limits

    At launch

    The 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.

See full pricing

  • Focused agent or agentic workflow
    Fixed price per phase
    From $25,000
  • Larger AI platforms
    Several agents, many integrations or a full product
    $100,000 and up
  • Scoping call
    Ends with a written scope and a price
    $0
  • Model usage
    Billed to your own account
    Paid 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

CTO and technology partner

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

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.

Or send the details

Book a 1-hour consultationWhatsApp us