From direction to working solutions.
GoSchool connects three parts of the offer: institutional decisions and readiness; a working platform, pilots, and adoption; and the integration or custom development the job requires.
Direction and readiness
Assessment, use-case prioritisation, AI strategy, governance, and role-specific workshops or training for leaders, educators, staff, and IT turn ambitions into shared institutional decisions.
Platform, pilots, and adoption
A measurable pilot can use GoSchool's working education product or a solution designed for the institution, with the required access, integrations, and measurement. What proves valuable can expand further; commissioned research using the institution's own anonymised usage data can also form part of the work.
Integration and custom development
Where the job requires it, we design and build API or MCP connections, administrative workflows, custom web interfaces, or institution-branded mobile applications.
We build the right solution for each task.
In some settings, GoSchool's working education product is the right choice. Others need institutional knowledge, process support, or system connections. We distinguish what is proven from what must be designed with the institution.
AI in university teaching
Students can ask questions, request explanations, and practise using materials selected by the educator. The educator defines the sources and boundaries.
This is where GoSchool began — in real use by educators and students since 2023.
Institutional knowledge and services
We design support grounded in authorised policies, documents, and other knowledge sources to provide useful answers and guide people to the appropriate institutional service. Actions inside a system always depend on the project and permissions.
A project designed around the institution's sources, roles, and specific task.
Processes and system connections
AI can gather information, prepare work, or connect to existing systems within permitted actions and human approval points.
Delivered through integration or custom development, as the use case requires.
The Platform page explains the technical model, permissions, and integration direction. Platform
Three years of learning what makes AI useful in practice.
GoSchool's first systems went live in university teaching in 2023. Since then, we have learned from real use by educators and students.
We bring this experience to wider institutional projects as well.
What three years taught us
Course knowledge makes the chat valuable.
In 2026, course-specific knowledge creates the value.
In 2023 it worked because the student got a good AI. Today the off-the-shelf models are stronger and more convenient, students go there — and they are right to.
What is leftThe course requirements, the standard for a good answer and the group's sticking points are local knowledge. A general model can learn them only from the course context.
Real use shows where to go next.
A computer graphics lecturer set weekly assignments that required continuous work. He suspected they were being done with AI.
The logs showed where students got stuck, what they asked, and what the AI supplied in place of understanding. He then configured the assistant to require an explanation before giving an answer. He was responding to a visible pattern of use.
PrincipleSeeing comes before the idea. Observable use provides the first foothold.
The most common pattern is frictionless shortcutting.
Outsourcing is the default. The largest pattern is a student asking for a concept, receiving the explanation, and stopping there.
Genuine dialogue — where the student's own thinking enters and is tested — appeared across different courses and different lecturers, with no common design feature behind it.
- 28%
- ask for a finished deliverable: an essay, a calculation
- 42%
- ask, receive, stop
- 20%
- multi-turn, but the synthesis stays with the AI
- 4%
- genuine dialogue, with the student's own thinking
Distribution of coded reasoning chains across 40,000 student messages, 15 institutions, 3 countries. Á. Tamási: The Power of Outsourcing (forthcoming).
What this meansWe treat outsourcing as a pattern of use. Course design explained the differences between the patterns we observed.
Course design guides how AI is used.
Where it worked, the lecturer had worked out what is worth teaching, and how it is worth assessing, when a model is in the student's hands.
In a group preparing for a statistics entrance exam, the assessment itself demanded understanding, so students outsourced towards understanding. The graphics course's weekly assignments predate AI and were set for pedagogical reasons. Both earlier teaching decisions shaped how AI was later used.
What followsAI use follows from a course-design decision. On a teaching task, the subject provides the starting point.
And the answer is often smaller than you expect.
On a Databases course, working through the exam questions is individual work on paper. In practice one student worked them out and the rest studied from that copy. The lecturer changed the arrangement so that each student worked through them with the AI — and interpretive questions appeared in the logs.
Outsourcing was already built into the course through the practice of copying.
In practiceOften, replacing an activity that lost its purpose long ago creates the change.
The open question
Where the task itself becomes outsourceable.
On the same graphics course, the logs expose a problem that remains open: where the core task — writing code — becomes entirely outsourceable, students hand in working output without understanding it.
This opens the next question: how do you teach a subject when output and comprehension come apart? The data already helps us ask it precisely.
Change connects people, operations and technology.
Leaders need shared direction, educators and staff need usable practices, IT needs clear boundaries, and the institution needs responsible operation — regardless of which technology delivers it.
Platform-independent work may include assessment, AI strategy, governance design, role-specific workshops or training, internal AI-champion enablement, and AI-aware course development. The outcome works with whichever technology the institution selects.
Together, we turn the goal into the right solution.
Show us the goal, obstacle, or initiative already in motion. Together we separate what calls for a professional service, the platform, integration, or custom development.