ES

DiaCroma™ for Education

Every student who quietly drops out takes their tuition with them.

How many of yours left over a hold nobody explained, a scholarship average nobody warned them about, or a requirement nobody told them had opened? The DiaCroma AI Tutor helps each student decide their next step — inside your university’s own app, in time to stay enrolled.

Calculate what it costs you Start with a conversation

In the U.S., 3 in 10 first-year students had left the institution where they started before their second year began.

National Student Clearinghouse Research Center · Persistence and Retention, fall 2024 cohort: 69.1% second-fall retention · source

The cost

What it costs you, with your own numbers.

Slide to your numbers

The starting values are only an example. Move each slider to match your institution.

Tuition that walks out every year

 

 

What keeping a few of them is worth

Keep 1 in 20
Keep 1 in 10
Keep 1 in 5

A scenario built from your own numbers, not a promise of results. The proof of concept measures the real effect in your institution, against success criteria agreed before it starts. And because the costly model never decides, guiding every student stays affordable.

Talk to us about these numbers

One student, two endings

Same student. Same hold. The only difference is whether someone told her in time.

Ana is ready to enroll for her next term. The system won’t let her. The deadline is August 15.

Without DiaCroma
  • Aug 2The portal says “hold on account.” No reason. No office.
  • Aug 5She writes to general services and is sent somewhere else. Twice.
  • Aug 12Still no answer. She stops trying.
  • Aug 16The deadline has passed. She is not enrolled this term.

Nobody decided to lose her. Nobody told her what to do.

With the AI Tutor
  • Aug 2She asks the app: “Why can’t I enroll?”
  • Same minuteIt names the hold — a pending course-load authorization — and who releases it.
  • Same minuteIt drafts her message to the coordinator and sets a reminder for August 13.
  • Before Aug 15She knows what is missing, who resolves it and by when.

One question. One step. In time.

A dashboard would flag her. A bot would quote the rule. The AI Tutor tells her what to do.

The DiaCroma app explaining to a student that a pending course-load authorization is blocking enrollment, drafting the message to the coordinator and setting a reminder.

The real app answering exactly this question (in Spanish). Ana is illustrative.

Why it keeps happening

Today’s tools are a bot or a dashboard. Neither gets the student to the next step.

BotDashboardAI Tutor
Talks with the student directly✓—✓
Knows this student’s live situation: record, holds, scholarship, balance—partial · no input from the student✓
Helps the student make the best decision toward their goal——✓
Cites a source for every answer, with a record you can defendpartial—✓

Each tool as it works on its own. A bot answers and stops; a dashboard flags a student and waits for someone to act. That follow-up does not always come.

What your institution learns

Save one student today. Learn where to change for all of them.

What could a school finally see if it could tell which students are quietly slipping, before they’re gone? Every conversation shows where students get stuck, which groups fall behind their goal and how each decision was made — so your leadership fixes the cause, not one case at a time.

The institutional view of DiaCroma: reach, re-enrollment against the prior period, after-hours share, each group against its goal, decision modes, cohort trajectories, activity and period health. Sample data.
The institutional view. Sample data.
Where students get stuckThe holds, offices and requirements behind the questions, campus by campus.
Each group against its goalNew students, at risk, scholarship holders, close to graduating.
Re-enrollment, attributableAgainst the prior period and measured conservatively: a number you can take to your board.
Every decision, on recordAnonymized, so any case can be reviewed.

From insight to change: the veritAI diagnostic →

DiaCroma™ for Education · AI Tutor

The DiaCroma AI Tutor.

The engine behind every student who stays: it understands the person and helps them make the best next decision — with published science behind every step.

See how it decides The research

The voice is the commodity. The decider is the thing no one else has.

What sets it apart · DiaCroma AI Tutor

The tutoring gains, without the tutoring risks.

In a randomized trial at Harvard, students with an AI tutor learned more than twice as much as in an active-learning class, in less time.
Kestin, Miller, Klales, Milbourne & Ponti · Scientific Reports 15, 17458 (2025) · doi:10.1038/s41598-025-97652-6

The authors are clear that the result depended on design: instructors wrote step-by-step solutions into the model’s instructions to keep it accurate, and they warn that hallucinations and over-reliance remain real risks. DiaCroma AI Tutor builds those safeguards into the architecture instead of the prompt.

  1. 1UnderstandLanguage models

    Read the student’s question and turn it into a precise request.

  2. 2DecideThe DiaCroma engine

    Machine learning to understand the student, code to hold the limits. The language models never decide.

  3. 3SayLanguage model

    Puts the decision into clear words. Nothing more.

The AI tutor in the studyDiaCroma AI Tutor
Where accuracy comes fromStep-by-step solutions written into the model’s promptDecisions made by the engine against your verified course material
What the language model decidesThe tutoring response, guided by the promptNothing. It reads the question and words the answer
The risks the authors flagHallucinations and over-relianceThe decision cannot drift or flatter, and every step is on the record

Comparison based on the design described in the published study.

How it decides

Four pillars, four gates, one answer.

What keeps a student enrolled is the decision behind the words. The AI Tutor relies on four pillars to decide:

1

The student, over time

What they are trying to do, where they left off, and the real limits they live inside: time, money, energy.

2

Your rules

Verified material, regulations, programs, deadlines and offices.

3

Your live systems

Under a strict order of truth: a live record beats a document, which beats a claim.

4

What works

Curated best practices for students in this situation — a guide, never an override.

Before it chooses, it removes what isn’t allowed for that student. Four hard gates, every time: feasible, safe, legal, legitimate. The wrong move never even competes. Then one of four answers:

Act

Take the next step

Explain, give a hint, move the student forward.

Ask

Ask what it needs

Only what it needs to decide well.

Options

Show the choices

When the trade-off is genuinely open.

Honest no

Or a person

If it doesn’t know or shouldn’t act, it says so, or hands the case to your staff.

The science

What the student experiences. And the research behind it.

What the student experiencesWhy it keeps them enrolledResearch
It never proposes something they can’t or shouldn’t doNo advice that lands them in a bigger problemAdmissibility-First Decision Control · The Four Admissibility Gates
It works on what is really holding them back, instead of asking for “more effort”The real obstacle gets fixed: capacity, willingness or coordinationThe Multiplicative Execution Bottleneck
It speaks to this student, with their time, money and energy — not to an averageA plan they can actually carry outTemporal Feasibility and Causal Credit
It looks after steady habits, not bursts of effortProgress that lasts through a hard semesterHabit Stability
When it isn’t sure, it says so and asksTrust: it never invents an answer about their futureHonest Belief Under Imperfect Evidence · Decision Honesty Under Uncertainty · Typed Decision Output
It learns each student without learning its way past the limitsBetter every term, never bolderThe Learning Boundary · Human Constraints as Mathematics
It measures what really worked, without claiming what would have happened anywayA retention number your board can trustTemporal Feasibility and Causal Credit
“Who decided you didn’t get the scholarship: a person, or a score nobody can explain?”

That is the first of fifteen questions in the DiaCroma manifesto. Ten research papers stand behind the answers, and the method is protected by U.S. Patent Application No. 19/640,556.

The manifesto →The research papers →

The result that matters: the student who makes it to graduation.

DiaCroma™ for Education · Beyond

All year long. Not only when something breaks.

Retention is where it starts. What follows: the student’s whole journey, the AI agents across your campus, and each learner, for life.

Any hour, in your university’s name. One answer, one next step.

Every moment of the student’s path

Six journeys, one AI Tutor. The right step at every moment.

Each situation on a student’s path has its own journey. It reads observable evidence — never a risk score — and takes the one next step toward the student’s goal, inside your rules.

New students

Unblocker

Finds the hold that is blocking enrollment and routes it to the office that owns it, so the student sees exactly what is missing and who resolves it.

holds · office directory
Scholarship holders

Scholarship & payments guardian

Warns the student before the next payment that puts the scholarship at stake — only if they consented. Without consent, it does not send.

balances & aid · consent first
Academic risk

Recovery

With two or more independent signals, it proposes one concrete recovery step and cites the evidence. With one weak signal, it asks before acting.

learning activity · evidence, never a score
Students who went quiet

Re-engagement

When silence crosses a threshold, it sends one warm message — and does not repeat it inside its window.

silence threshold · one nudge per window
Close to graduating

Graduation

Names the next missing document against its deadline. When two paths are both good — the fastest, or the one with the nearest deadline — it shows both.

documents · deadlines
On track, with friction

Employability

Tells the student when a requirement opens — in Mexico, social service once 70% of credits are earned — something the system can compute and nobody tells them.

student record · informs and routes

For the student, every day

What is pending, and who to turn to.

In one place

Everything pending

Enrollment, scholarship, social service, courses, graduation: each student sees what is open and what comes next.

It keeps count

Hours, credits and dates

With a warning before each deadline.

The app telling a student they have 180 of 480 social service hours, that at their pace they finish in March, and that it will warn before each pending report.

Around the clock

And your staff, when it matters

Answers at any hour from your regulations, calendars and offices, giving your staff their hours back. When a person is needed, the case arrives with full context, not a cold ticket.

Beyond the student app

Your campus is already building AI agents. DiaCroma governs those too.

Agents built in Microsoft Copilot Studio, in Foundry or in your own code can run under DiaCroma for Agents. What each agent may do is checked against everything it has already done, before it acts — the same discipline that protects your students, applied to every agent on campus.

Copilot StudioFoundryYour own codeAvailable now on Microsoft Marketplace

DiaCroma for Agents →

Your own DiaCroma

Your knowledge, your rules, your brand. And each learner, for life.

Adults of different ages back in a classroom.

Your institution’s version

Your knowledge base, your rules and limits, your offices and your brand, owned and maintained by your institution. With a clear agreement on who does what: data ownership, security, operations, human escalation, reporting and continuous improvement.

Lifelong learning

Fewer traditional students, degrees that carry less weight on their own, and AI changing what is worth learning. The institutions that come out ahead will serve each person for life: as students, as alumni, as professionals who come back. DiaCroma knows each learner over time.

DiaCroma™ for Education · How to start

Two ways to start. Both of them small.

With your students, through DiaCroma for Education, or with your institution, through veritAI. You choose where to begin.

A university leadership team working together around a table. DiaCroma for Education

Start with your students

A proof of concept · a defined group · in your environment

  1. A first conversation. Your size, the systems you want to connect and who will run it. We reply with a proposal.
  2. We agree the scope. Student group, campus, priority journeys, channels, language, action limits, human escalations and success metrics.
  3. We install it in your environment. Your identity and one or two systems: SIS/ERP, payments, scholarships, LMS or documents. The proposal states what is integrated, what is simulated and what stays out.
  4. We measure and decide together. Decision records, student outcomes, escalations and workload, against the criteria agreed at the start. Then campus by campus.
Request a proposal
A yellow staircase seen from below. veritAI

Start with your institution

A 30–60 min scoping session · then the 90-day diagnostic

  1. A scoping session. With one executive, no obligation: a first estimate of your AI maturity, the AI already in use without governance, and an honest answer on whether the diagnostic is the right next step.
  2. Where you stand. A 10-dimension maturity matrix per department, with your risk appetite determined, and an audit of your technology stack.
  3. Who moves it. A profile of each leader — capacity, willingness and coordination — with an action plan per person.
  4. How it gets funded. A financial model built from your actual metrics and a first phase with owners and governance gates. On Day 90, all of it in a live decision-support system.
Request a scoping session

veritAI · the first 90 days

DiaCroma shows you where to change. veritAI helps you make the change.

veritAI is Verit’s transformation practice. We don’t ask you to stop running your institution: sessions of 60–90 minutes per leader per week, pre-scheduled, each with a defined output. We prepare everything; you receive outputs, not homework.

Weeks 1–2

Immersion

Leadership interviews, an inventory of ungoverned AI and a review of your systems. First findings by week 2.

Weeks 3–6

Level and risk appetite

The maturity matrix per department, your appropriate risk level, each leader’s profile and benchmarking against institutions like yours.

Weeks 7–10

The first investment

Specific initiatives with owners, governance gates and measurable outcomes, and a return your CFO can defend.

Day 90

Handoff

The four outputs, assembled into one system and in use. From Day 91, a critic agent improves it with every query.

Built for higher education

Flagship research or tuition-driven private; COPAES, FIMPES, SACSCOC and QS accreditation; the faculty adoption curve; student continuity. veritAI’s maturity framework has eight levels; its universities edition is coming soon:

LevelIn a university
L0 · GovernanceProtects compliance and institutional accreditation.
L1 · Augmented departmentsStudent-success AI and early-alert models.
L2 · Data foundationUnifies enrollment, registrar, LMS and advising data.
L4 · Role-aware surfacesProvosts, deans, advisors and registrars, each with their own decision context.
L7 · Portfolio intelligenceEnrollment competition, labor-demand signals and alumni-network intelligence.
The first 90 days →The L0–L7 maturity framework →

DiaCroma™ for Education · Questions

What your team will ask before saying yes.

How it fits your university, where your data lives, and what it does in the hard cases.

How it fits your university

You already have the pieces. None of them acts on its own.

In · yours

Your information

  • Regulations, calendars, services and offices
  • Student records, scholarships, payments, the learning platform
In the middle · ours

The AI Tutor

Out · yours

Your channels

  • Your app, your sign-on, your brand
  • Your offices, when a person is needed

Your information and your channels stay yours. We bring the decision.

IT and legal

The questions, answered.

Does an AI model decide what happens to our students?

No. The decision belongs to the AI Tutor, inside limits it cannot cross; the language model decides nothing. See how.

Where does our student data live?

In your institution’s environment, behind your single sign-on, from day one. If you prefer, Verit operates it for you.

Can we show what it did, and why?

Yes. It is designed so every decision is written to a record your team can replay: what was weighed, what was set aside, and why.

What happens when a student is in crisis?

It does not counsel. It routes to your wellbeing team, and it never follows up unless the student does.

Will it message students without their permission?

Scholarship and payment warnings go only to students who consented; without consent, nothing is sent.

Do we have to connect every system first?

No. You start small: see How to start.