Eudemonic AI · Anonymous · Cohort-level · Human-in-the-loop
A strong placement record does not mean students are at ease
Eudemonic AI shows institutions the stress that sits underneath the numbers they already know — anonymously, at the level of a cohort, before it becomes a complaint, a dropout, or a crisis.
Request a pilot See how it works
The case
Why a strong placement record still leaves a Dean blind
Placement numbers measure an outcome. They do not measure the load behind it.
A student with a ₹1 crore offer can still be anxious — about company expectations, about whether they can meet them, about whether they will have to resign, and about leaving home, their state, or the country. The offer did not remove the anxiety. It changed its shape. And because the number on the placement report reads as a success, nobody goes looking.
A student without an offer needs something else: a mindset that can hold alternatives — a startup, a business, other employment, higher education and then work. The job matters. The total mindset matters more, and that is the thing an institution can actually build.
Stress in a cohort is not one thing. Placement pressure is external. Peer pressure is social. Financial stress is personal. Personal wellbeing sits in the present: the student who looks fine on a marksheet and is quietly burning out. A topper can carry more anxiety than a struggling student, because staying on top is not the same pressure as catching up.
Why the student is not named
If the Dean can identify the student, the survey stops being honest. Anonymity is not a feature withheld from the institute. It is the reason the data is real.
What the institute gets is the pattern: this share of a year, in this programme, reporting financial strain, relocation fear, or anxiety even among those with offers. That is enough to run a workshop, a counselling drive, or a session on alternatives — without singling anyone out.
And when something goes wrong, the question is always “did you know?” Cohort-level insight is how the institute answers that before the incident, not after it.
What the pattern is made of
Stress in a cohort is not one thing
Treating it as one thing is why generic wellbeing initiatives so often miss. These four pressures behave differently, surface differently, and need different responses.
Placement pressure
External. Driven by the market, the recruiter list and the comparison with last year's cohort. Largely outside the student's control, which is precisely what makes it heavy.
Peer pressure
Social. Who got what, who got it first, and what that is taken to mean. It intensifies in exactly the weeks an institution is celebrating.
Financial stress
Personal. Fees, loans, family expectation and the cost of relocating. Rarely spoken about on campus, and almost never raised with a department.
Personal wellbeing
Present tense. The student who looks fine on a marksheet and is quietly burning out right now.
The topper is not the safe case
A topper can carry more anxiety than a struggling student. Staying on top is not the same pressure as catching up — it has no finish line, and the student has further to fall. Academic performance is a poor proxy for how a student is doing, which is why looking at results and concluding the cohort is fine is a mistake institutions make routinely.
The design decision that matters most
Why we will not tell you which student it is
Institutions ask for this, reasonably: if a student is struggling, surely the quickest help is to know who.
If the Dean can identify the student, the survey stops being honest. Students answer a named survey the way they think it should be answered. They understate financial strain. They do not admit to doubting an offer they are supposed to be grateful for. The data becomes a record of what students believe the institution wants to hear.
Anonymity is not a feature withheld from the institution. It is the reason the data is real.
What the institution gets instead is the pattern. This share of this year group, in this programme, reporting financial strain. This share reporting relocation fear. This share reporting anxiety even among those holding offers.
That is enough to act on. It is enough to run a workshop on alternatives, open a counselling drive before the placement season rather than after it, bring in a session on managing a first corporate role, or change how the department talks about offers in a year when the market is thin.
None of that requires singling anyone out. All of it is harder to justify, and easier to delay, without the number in front of you.
What Eudemonic AI is — and what it is not
Eudemonic AI is a decision-support and insight tool. It is not a diagnostic or automated safeguarding system.
It does not diagnose mental illness. It does not make automated decisions about individual students. It is not a substitute for your counselling service, your safeguarding arrangements or your crisis response — it is intended to inform them. Outputs are designed for cohort-level staff review and require human interpretation before any institutional action is considered.
The platform
Six capabilities behind the pattern
Research-informed questionnaires
Structured questionnaire workflows capturing academic stress, workload perception, support access, belonging and related themes in a format the institution can use.
Cohort-level analytics
Anonymised, aggregated analysis of patterns across student groups, programmes, demographics and successive survey waves.
Machine-learning-assisted insight
Clustering and structured thematic analysis surface patterns across both the numeric answers and the free text, for staff to read and interpret.
Earlier pattern visibility
Emerging pressure themes become visible earlier than an end-of-module or annual feedback cycle typically allows.
Readable dashboards and summaries
Interpretable dashboards and structured summaries built for a committee conversation, not for a data team to decode first.
Human-in-the-loop by design
Every output supports staff interpretation and professional judgement. The platform informs the review; it does not replace the people making the decision.
How it works
From questionnaire response to institutional insight, in five steps
Questionnaire delivery
Students complete a structured questionnaire designed for honest reflection on academic stress and related wellbeing themes.
Analytics processing
Responses are processed through analytics and machine-learning-assisted workflows.
Pattern surfacing
Cohort-level themes, stress clusters and areas for review are identified in anonymised, aggregated form.
Dashboard and reporting
The institution receives dashboards and summary outputs for staff-led discussion and planning.
Evidence-based planning
Staff use the insight to prioritise workshops, counselling capacity and targeted institutional review.
Throughout: your judgement
At no point does the platform act on a student. Each step produces material for people to read, question and decide on.
The mentoring module
A list, a log, a button and a card
Analytics tell an institution about a cohort. They do not help the faculty member who has twenty mentees and fifteen minutes. The mentoring module is the other half of the platform — and it is deliberately built so that a mentor is never turned into a psychologist.
A list
Their own mentees, drawn from the roster, each showing a first-contact date or not yet met. When a mentor changes, the handover carries the relationship rather than resetting it.
A log
A contact record that takes about thirty seconds on a phone, private to the mentor by default. The same log serves as the anti-ragging diary and as accreditation evidence, without being written twice.
A button
One request support action that routes to the right door, carries the minimum dataset, and records urgency and the student's agreement.
A card
Get help now — on every mentor screen, one tap, bypassing every queue. It works offline, exists as a printable card, and as a public page carrying no personal data.
Prompts in words, never scores
The prompts a mentor sees are sentences, not alerts: not yet met, no contact lately, asked to talk, missed meetings, follow-up due. Attendance and marks prompts begin only when the institution connects its ERP data and sets its own thresholds, and they pause during protected absences.
No mentor screen, export or notification ever shows a wellbeing score, a band, a rank, a screening result or a diagnosis. There is no machine learning anywhere in the mentor's path. A prompt derived from survey answers is sent only where the student has agreed to it, only from sections the institution has marked non-clinical, and only in supportive language.
The right door, and an answer back
Requests route to counselling, the wellbeing office, studies, the health centre, money and scholarships, disability, careers and the hostel — and to the statutory bodies: the anti-ragging committee, the grievance committee, the Internal Complaints Committee and the Equal Opportunity Cell. Each door has its own rule for what is sent and whose agreement is needed.
Status comes back. Received always; anything beyond that only with the student's agreement; and a request left unacknowledged past its door's time limit escalates to a named owner rather than going quiet. For statutory bodies the mentor's record keeps only signposted to [body].
Separation: what the mentor records stays where it belongs
Students tell a mentor things only if those things do not travel. Nothing a mentor records reaches marks, discipline, parents, name lists or accreditation files without the student's agreement or a legal duty — and that is enforced by automated tests, not by policy alone: no report, export, email, webhook, API response or audit entry carries mentor note text.
The exceptions are the ones the law does not leave to anyone's discretion: child sexual abuse and danger to life are legal-duty routes that do not wait for agreement.
Training register and role statement
Who is trained, by which provider, whether that provider is a certified mental-health professional, hours, date and next due. The institution's role statement appears on the log form itself, where the boundary actually matters.
Counts, not league tables
Coverage, time to first meeting, acknowledgement and closure times, training coverage — by department or programme, with small numbers suppressed. Mentors are never ranked and student outcomes are never shown by mentor.
Statutory returns as a by-product
Counts for NBA criterion 9.2, the NMC monthly report and AICTE monthly complaint returns come out of work already done, rather than being assembled separately when the return falls due.
Regulator presets
Referral, confidentiality, follow-up, record-keeping and allocation rules are institution settings with cautious defaults and presets per regulator — UGC, AICTE and NMC — because these requirements are still moving.
Students under 18
A guardian's consent is required, monitoring is held back without it, and minors are excluded from analytics — as the DPDP Act 2023 requires for children's data.
Accessible, and in Hindi
The pages students, guardians and visitors use are checked against WCAG 2.2 AA with a published accessibility statement. The student page, the survey and the questionnaire translation sheet support Hindi.
What the mentoring module deliberately does not do
- Diagnose, label, or show clinical information to mentors.
- Score, rank, cluster or predict students — there is no machine learning in the mentor's path at all.
- Show survey answers to mentors.
- Rank mentors, or show student outcomes by mentor.
- Show anything to the parents of adult students.
- Export mentor notes anywhere.
- Monitor students continuously — no location, Wi-Fi, hostel logs or social media, and no scanning of what they write.
- Talk to students about their feelings through AI.
- Put an emergency in a queue, or ask a mentor to judge risk.
- Investigate complaints, or keep accounts of alleged events in mentoring records.
- Call a mentor a counsellor.
Who uses the output
Built for the people who answer for it
Dean of Students, Director of Student Affairs
Where pressure is concentrating across programmes and year groups, early enough to plan a response rather than account for one afterwards.
Placement and career services
What the placement report cannot show: how the cohort is actually carrying the season, including the students the numbers count as successes.
Counselling and pastoral teams
Evidence for where to direct finite counselling capacity, rather than allocating it to whoever happens to present.
Registrar and academic leadership
Workload perception and transition-related concern alongside the academic record, informing timetable, assessment and induction decisions.
Quality and accreditation teams
Documented, repeatable evidence of student voice and wellbeing review — the kind of material NAAC, NBA and institutional quality frameworks ask to see.
Directors and governing bodies
Summary outputs for board discussion, showing both what the data indicates and the limits of what it can support.
The question afterwards
“Did you know?”
When something goes wrong at an institution — a withdrawal, a complaint, an incident, an inquiry — the question put to the leadership is almost always the same one. Did you know? Was there anything that should have told you?
Cohort-level insight is how an institution answers that before the incident rather than after it. Not because it predicts who, but because it shows that the institution was looking, found a pattern, and acted on it — with a record of when and what was done.
Governance
Data protection and professional boundaries
Wellbeing data is among the most sensitive information an institution holds, and the platform is built on that assumption. Analysis is anonymised and aggregated at cohort level; outputs are for staff review, not automated action.
The institution remains the controller of its own data — the Data Fiduciary under India's DPDP Act 2023, or the controller under the UK or EU GDPR where those apply. Processing is governed by the implementation contract and a data processing agreement.
Alliance and delivery
Who builds it, who implements it
Eudemonic AI is a product of OLT ERP Limited, a London-based EdTech company registered in England and Wales, at 103 Russell Lane, London N20 0AZ. It was developed in collaboration with UK universities.
A & S Software Consultancy Pvt. Ltd. is OLT ERP Limited's alliance partner and delivers implementation, integration and support for institutions in India — including alongside an existing OltErp deployment, where the two run as separate systems serving different purposes.
The two companies are separate legal entities. Each contracts in its own name and is responsible for its own obligations. Which entity you contract with will be stated clearly in any proposal.
Track record
Where it has been used
Figures published by OLT ERP Limited at olterp.co.uk.
Explore a pilot for your institution
One survey wave, one cohort. See what the pattern looks like in your context before committing to anything wider.