AI that reads the institution — not the open web
A retrieval-augmented assistant grounded in your pathway templates, SIWES cycles, employer briefs, and counsellor playbooks. Counsellor-supervised by default.
An assistant that actually knows your campus.
Every answer cites its source. Every session can convert to a counsellor case with full context. Every row of data stays partitioned to your institution.
Four capabilities, grounded on institutional data
Each capability reads a structured, NDPR-filtered subset of your institution's records — never the open web.
Pathway coaching
Explains Holland RIASEC and Big Five results in the language of the student's faculty, then proposes milestone sequencing that fits the SIWES calendar.
Application drafting
Drafts CVs, cover letters, and statements of purpose using validated skills, transcripts, and completed pathway deliverables — never fabricated.
Skills diagnostics
Interprets digital, numerical, and domain assessments, recommends targeted learning modules, and tracks the gap closing over time.
Counsellor prep
Produces a 5-line case brief before every appointment: stage, risk, last touch, next action, suggested talking points.
Compliance, by design
Built on NDPR obligations and the Nigerian education data landscape.
NDPR-grounded by default
Student data stays in-region, in tenant-isolated partitions. AI prompts and responses pass through an NDPR filter that strips or redacts before logging.
Counsellor-reviewed escalations
Any response touching mental health, welfare, disability, or finance routes to a human before reaching the student.
Audit trail per institution
Every prompt, retrieval, and response is signed, timestamped, and linked to the active user. Institutions can export the full trail on demand.
No open-web grounding
Retrieval only pulls from your pathway templates, handbooks, partner briefs, and approved reference material. The open web is explicitly out of scope.
Sample prompts students actually send
Drawn from anonymised UNILORIN and OAU logs — paraphrased for publication.
“Which SIWES placements fit my Holland profile and keep me in Lagos?”
“Help me draft a cover letter for the Flutterwave fellowship using my validated TypeScript and system design work.”
“My pathway is at 41%. What should I prioritise before the end of this term?”
“Compare my skills against Access Bank's Management Trainee role and flag the two biggest gaps.”
“Summarise my last three counselling sessions and tell me what Dr. Mariam wants from me this week.”
Data isolation at the tenant boundary
Each institution gets its own knowledge partition and its own retrieval index.
How a prompt is served
Captured with tenant id, role, matric and session id. NDPR filter strips regulated fields before logging.
Scoped to the institution's partition: pathway templates, handbooks, SIWES calendars, approved employer briefs.
Retrieved passages form the grounding context. Model cannot reach the open web. Output includes citations.
Response passes safety + NDPR post-check. Sensitive topics route to counsellor queue before reaching student.
Student sees response with citations. Counsellor can convert to a case. Audit log records prompt, retrieval, response.
Deploy an AI assistant your DPO will sign off on.
Tenant-isolated retrieval, NDPR filters, and counsellor review for every sensitive topic.