Institutional governance
Faculty set the boundaries. The institution sets the policy.
Mika deploys inside your governance, not around it. Roles and permissions, per-course policy controls, reporting that rolls up from a single course to the whole institution, and a record of who changed what — so a chair, a dean, and IT can each see exactly what they are accountable for.
Compliance
Meets the rules your institution answers to.
The privacy, accessibility and AI rules universities are held to in the US, UK and EU, and what each one means inside Mika. Every line below describes how the product works today.
EU · UK
GDPR
EU GDPR · UK GDPR · Data Protection Act 2018
Hosted in the EU: database, files, servers and AI processing run on Google Cloud in Belgium and the EU region. The institution is the controller and MikaLabs its processor under a DPA. Students download or delete their own data from Settings, and the few US services involved are covered by the Data Privacy Framework (with its UK Extension) or Standard Contractual Clauses.
US
FERPA
Family Educational Rights and Privacy Act
MikaLabs acts as a school official. Education records are used only for the institution’s purposes and under its control, and never passed on except as it directs or the law requires. Inspection and correction requests run through the institution, and we help it respond.
US · UK · EU
Accessibility
ADA Title II · Section 508 · EU and UK public-sector rules · Equality Act 2010
Mika’s main screens are built to WCAG 2.1 AA, the standard these laws point to. Teachers can set exam adjustments (extra time, no time limit, or an assistive-technology exemption from integrity checks), and only the adjustment is stored, never the reason for it.
EU
EU AI Act
Transparency and human oversight
Students are asked before any of their content reaches an AI model, and can change their answer at any time. AI-judged marks stay provisional until a teacher approves them, any student can ask for a human review of a mark, and no student data is used to train AI models.
The oversight model
How Mika sits inside an institution.
01Four views, one deployment
Mika runs four views: student, teacher, course coordinator, and administration — the last for whoever owns the level above the course, such as a dean of science. Each view sees the layer it is responsible for and the aggregate beneath it.
02Reporting that rolls up
A teacher sees their course, a course coordinator sees the course across its sections, and administration sees the faculty above it — the same underlying data, aggregated to the level each view is accountable for.
03Human in the loop by default
Mika Consultant drafts lesson plans, assignments and student outreach; a person releases them. The institution decides which agent actions, if any, are allowed to run unattended.
04Deployed without an IT project
Identity, SSO and SIS integration are configured by MikaLabs during deployment. Your institution does not need to staff an integration team to stand Mika up, or to keep it running.
Who sees what
One deployment, four vantage points.
Access is scoped to accountability. Each view sees the layer it is responsible for and the aggregate beneath it — never the layer above.
| Capability | Student | Teacher | Course coordinator | Administration |
|---|---|---|---|---|
| Own progress and learning historyWhat a student sees about themselves | full | none | none | none |
| Course analytics and gap reportsPer-concept attainment across a cohort | none | full | full | none |
| Cross-course and section rollupAggregated across every section of a course | none | none | full | full |
| Faculty-wide reportingEvery STEM course in the faculty under one view | none | none | none | full |
| Curriculum and rubric configurationWhat Mika is aligned to teach | none | full | full | none |
| AI policy and content-filter controlsWhich capabilities are enabled, and how far they go | none | none | full | full |
full — complete access at that levelscoped — aggregate or own-record onlynone — not visible in this view
Academic integrity & AI policy
Guardrails you set, not guardrails we assume.
Policy is set per course
Each course carries its own policy: which Mika capabilities are available, how much the tutor may reveal, and what changes during an assessment period. Faculty set it, within limits the institution defines.
Withholding is the default, not a setting
Mika is built to lead a student to an answer rather than hand it over — a pedagogical decision documented in our published research, not a compliance toggle added afterwards.
Content filters under your policies
Filter controls run under the institution's policies, not ours — the same posture that governs curriculum alignment and usage analytics.
Flagging and escalation
Gap and misconception detection identifies students falling behind, and Mika Consultant can raise it. Whether that reaches the student directly or routes to a human first is the institution's decision.
Program review & accreditation
Evidence that already exists when review comes around.
Outcome-aligned by construction
Mika ingests your syllabus, course materials and assessment rubrics, so the mapping between what is taught and what is measured exists from day one rather than being reconstructed at review time.
Per-concept attainment evidence
Deep Analytics reports attainment concept by concept across a cohort, and gap detection traces a wrong answer to the missing prerequisite behind it — evidence at the grain a program review actually asks for.
Exportable for review cycles
Course, program and institution reporting can be exported for accreditation submissions, program review, and internal quality cycles.
Security, data & procurement
The answers your review board asks for.
Views
Four — student, teacher, course coordinator, administration
Identity & SSO
SAML/OIDC and SIS integration, configured by MikaLabs on the institution’s behalf
IT lift
None — setup is run by us, not by your IT department
Training on student data
Never — contractually guaranteed, not a default toggle
Procurement pack
Security questionnaires, DPAs and reference architectures on request
Architecture
Full threat model and data-flow whitepaper under NDA