The professor marks their own homework
The person who taught the unit decides whether the unit was learned. Even with total integrity, that is a conflict of interest built into the timetable.
How we're staffed
Every promise a university makes needs a person whose whole job is to keep it.
This is the disaggregated faculty model: pioneered at Western Governors University across hundreds of thousands of students, adopted and advanced by Forward, and enforced end to end inside Amagezi.
The problem, honestly
Think of the best professor you ever had. One person, six jobs.
They designed the course, delivered it, mentored you, marked you, published research and ran the administration. When it worked, it was extraordinary. It failed in five predictable ways.
Mentoring is nobody's actual job
It is what a good academic does after teaching, marking, designing, researching and administering. It is the first thing a busy term deletes.
Quality depends on one person's bandwidth
A brilliant lecturer with a sick child, a grant deadline or four extra tutorials becomes, for that month, a worse teacher. The student pays for it.
It cannot scale without diluting
Doubling students means doubling generalists, each stretched the same way. Quality is diluted by arithmetic, not by intent.
It cannot deliver mastery-based progression
Progression on evidence rather than on the calendar needs marking capacity that is independent of teaching capacity. One person cannot be both clocks.
None of these criticise academics. Every one of them criticises a job description.
Grounded inForward University Faculty Model (founding document) the Vice-Chancellor's founding faculty design paper
The precedent
Western Governors University asked the question first.
Founded in 1997 by nineteen United States governors, WGU was built to answer one question: what would a university look like if it were designed around competency rather than the credit hour? Its answer reshaped its faculty, programme mentors, single-course instructors, independent evaluators and central curriculum teams, and the university scaled to hundreds of thousands of students without the model breaking.
What the model proved
- Mentoring became universal instead of accidental.
- Assessment became consistent and defensible, because markers were independent and rubric-bound.
- Courses became institutional assets rather than private property.
- Scale stopped diluting quality: disaggregation is a machine for producing it.
What Forward adds to it
- An AI teaching assistant inside every course, under named faculty supervision.
- A practice tutor role for the cooperative education WGU's model never needed.
- The human authority rule that governs the AI Oral Examination Room: no machine ever has the final word on a student.
Adopt what is proven; advance what is ours.
Grounded inForward University Faculty Model (founding document),Western Governors University, founded 1997 WGU's published history and our own design paper
Five principles govern the design.
- Disaggregation
- The professorship is divided into specialised full-time roles. No role wears two hats that conflict.
- Separation of teaching and evaluation
- The person who teaches a student never marks that student's assessment. Amagezi enforces the separation in software, not in memory.
- Named accountability
- Every student can name their mentor. Every course names its instructor and its architect. Every promise in the prospectus maps to a role.
- Mastery needs service levels
- Because students progress on demonstrated mastery, evaluation, feedback and guidance carry published turnaround times, tracked like any other promise.
- Humans decide, AI assists
- AI teaches tirelessly, drafts and flags; a named human confirms everything that counts, exactly as in the AI Oral Examination Room.
The model
One professor becomes a team of specialists.
Nothing is removed. The six jobs are separated so each one belongs to somebody whose entire job it is.
Traditional model
One person, stacked
- Design the curriculum
- Deliver the teaching
- Mentor the students
- Mark the assessments
- Research and publish
- Administer the programme
Quality is a function of one person's remaining hours.
The Forward model
Six specialists, one student
Your Programme Mentor
One named person, from enrolment to graduation.
Not a personal tutor you meet once. A named professional whose entire job is your progress: your pace, your pathway, your blockers, your entry dates, your co-op timing. They call you on a fixed cadence, and they call you when the data says something has changed.
The answer to the student nobody noticed drifting.
Your Course Instructors
Subject answers within one working day.
Specialists who do one thing: teach their subject to students who are actively stuck in it. They hold clinics, run live sessions and answer directly, because answering is the job, not the overflow from five other jobs.
A subject expert answers you within one working day.
Your Evaluators
They never taught you. That is the point.
Trained assessors who mark against the published rubric and nothing else. They do not know your seat, your effort or your excuses. They know your evidence. Marking is returned inside 72 hours.
The reason a Forward result is evidence, not opinion.
The Curriculum Architects
They design the unit; they never mark it.
Academics and industry practitioners who build competence units, rubrics and assessment instruments, then hand them to instructors to teach and evaluators to judge. Design quality stops being a lottery of who was assigned the module.
Every rubric is published before the work is set.
Your Practice Tutor
The paid-work promise, delivered by somebody.
Your co-op is not a form you file. A practice tutor prepares you for placement, sits between you and the employer while you are there, and assesses the work-integrated learning when it is done.
Twelve months of paid placement, with a named person accountable for it.
Musomesa, under named human supervision
Available at 11 p.m. Never the final authority.
Musomesa explains, drills, rehearses and scores first-pass attempts at any hour. Every one of its behaviours sits under a named academic supervisor, and every decision that counts is confirmed by a human. AI expands contact time; it never replaces academic authority.
A human answers for everything that counts.
The wall between coaching and judging
Coach side
Wants you to pass, and helps you get there.
- Programme Mentor
- Course Instructors
- Practice Tutor
- Musomesa, in practice mode
Judge side
Has no stake in the outcome, only in the truth.
- Evaluators who never taught you
- Assessment integrity review
- Independent re-mark panel
- Human confirmation of every AI score
Your teacher wants you to win. Your marker only wants the truth. Keeping them separate protects both.
A fortnight in the model
Amina never had to ask whose job it was.
Select a moment. Every one of them is a different specialist doing the only job they have.
Programme Mentor
The call that is already in the diary
Amina's mentor opens with her pace, not with small talk: two units cleared, one drifting, co-op window moving to the July block. They agree the week before it happens to her.
Monday, 09:40
At no point did Amina wonder whose job her problem was.
Grounded inForward University Faculty Model (founding document),Amagezi platform specification role definitions and the engines that schedule them
One course through the model
Machine Learning Foundations, from build to version 1.1.
Now watch the same machine from the course's point of view: four competence units, three hundred students this cycle, one standard.
- 01Curriculum Architects
Built once, centrally
Machine Learning Foundations, 4 competence units, is built from the programme's competency register: outcomes, content, activities, the assessment, its rubric, an oral question bank with three approved variants per slot, and the grounding pack for the AI assistant.
- 02Programme Chair
Approved as version 1.0
No course reaches a student until the academic owner of the programme signs the version.
- 03Course Instructor
Delivered in both pathways
Weekly sessions in the Term Model, an intensive fortnight in the Block Model, plus the course room, the clinics, and supervision of the AI assistant. Three hundred students take it this cycle.
- 04The student
Mastery demonstrated when ready
A project submission or the oral examination, the student's choice, when the evidence is ready rather than when the calendar says so.
- 05Evaluators
Marked blind, inside 72 hours
Evaluators mark against the rubric and confirm every AI provisional score. The instructor sees results only in aggregate; the evaluator sees the student only as evidence.
- 06Architects, then the Chair
Version 1.1
Criterion 3 confused a fifth of the cohort, so the architects add a worked example and the Chair signs it. The course improved without any student's result depending on who happened to teach them.
Four pairs of hands touch the course; one standard governs it; the marker never taught the student. That is what an institutional asset means.
Grounded inForward University Faculty Model (founding document),Assessment and quality framework course ownership and rubric publication rules
Who owns what
Every promise, its owner, and where it is measured.
| The promise to the student | The role that owns it | Where it's measured |
|---|---|---|
| You will never be anonymous here. | Contact cadence on the mentor board | |
| Someone who knows the subject will answer you. | First-response time, one working day | |
| Your result means something outside Forward. | Independent marking, blind to teaching | |
| You will be marked against a rubric you were shown first. | Rubric publication date on every unit | |
| You will do real, paid work before you graduate. | Placement and co-op assessment records | |
| Help exists at the hour you are actually stuck. | Round-the-clock availability logs | |
| A human answers for everything that counts. | The human authority rule, audited |
Select any role to see its exact responsibilities, measurement rules and hiring process.
Subject answer turnaround
1 working day
From a student question reaching a course instructor to a substantive reply.
Marking turnaround
72 hours
From submission to a returned evaluator decision against the published rubric.
Programme mentor calls
Every 2 weeks
Scheduled human contact with a named mentor, whatever a student's engagement level.
A university's real prospectus is not its brochure. It is its faculty model.
Grounded inAmagezi platform specification,Published service commitments published targets, enforced as timers
Caseloads, ratios and scale
Published ratios, because a caseload nobody measures is a promise nobody keeps.
These are the University's design targets for every role, alongside the service promise each ratio exists to protect.
| Role | Design caseload | Service promise | Scales by |
|---|---|---|---|
| Programme Mentor | 1 to 80 in year one; 1 to 120 at steady state | Contact every 2 weeks in first year; risk response in 2 working days | Hiring mentors per 120 enrolments |
| Course Instructor | 1 per 250 active students per course | Subject answers in 1 working day; clinics before every window | Adding instructors per course as cohorts grow |
| Evaluator | About 40 assessments per week each | Marking returned in 72 hours; calibration weekly | A marking pool sized by Amagezi's forecast |
| Curriculum Architect | Team of 3 per about 40 live courses | Course versions each cycle; fixes in 10 working days | One team per School, growing with the catalogue |
| Practice Tutor | 1 to 60 concurrent placements | Two visits per placement; issues acted on in 3 days | Tutors per placement volume, per region |
| AI Teaching Assistant | Unlimited students, one course | Instant, always on, always cited | Compute, under instructor supervision |
| Programme Chair | 1 per programme, all 91 | Boards on schedule; monthly dashboard review | Fixed by catalogue |
| Dean of School | 1 per School, all 10 | Staffing plan reviewed each term | Fixed by structure |
Enrolment growth translates into published hiring ratios, not silent overload. A breached caseloadCaseload: The number of students, courses or assessments a role is designed to carry, published, not private. Select to jump to the glossary. is a Dean's action item, visible on the same dashboards as everything else.
Grounded inForward University Faculty Model (founding document),Published service commitments design ratios and the promises they protect
Enforced in software
A faculty model that lives in a chart dies in the first busy term.
So this one does not live in a chart. The roles, the separations, the caseloads and the promises are objects Amagezi knows, enforces and reports, term in, term out, whether anyone is watching or not.
- Assignment engine
- Routes every assessment to an eligible evaluator, enforcing the separation ruleSeparation of teaching and evaluation: The rule that no one marks a student they taught, enforced by Amagezi's assignment engine. Select to jump to the glossary. and balancing load against the 72-hour promiseService promise: A published turnaround, such as marking in 72 hours, tracked by a timer in Amagezi. Select to jump to the glossary..
- Scheduling engine
- Mentor call cadences, clinics and boards, created as calendar objects both sides can see.
- Promise engine
- Every service levelService promise: A published turnaround, such as marking in 72 hours, tracked by a timer in Amagezi. Select to jump to the glossary. is a timer. Breaches notify the role holder, then the Chair, then the Dean.
- Staffing forecast
- Enrolment projections times the published ratiosCaseload: The number of students, courses or assessments a role is designed to carry, published, not private. Select to jump to the glossary. equal the hiring plan, per School, per term, automatically.
- Audit trail
- Every assignment, decision, override and escalation writes to one append-only trail.
A dashboard per role
- The Mentor's caseload board, with a nightly risk light per student.
- The Instructor's course room, with unanswered questions aged against the one-day promise.
- The Evaluator's blind marking queue, with the 72-hour clock visible.
- The Architect's studio, with per-criterion analytics and pathway equivalence checks.
- The Chair's and Dean's boards, with every promise as a tracked line: green or breached.
What makes the Forward version unique.
The AI teaching assistant, governed
A tireless assistant in every course, and then the harder thing: its human supervisor named on screen, grounded on the course, every answer logged, and banned from assessing.
Practice inside the faculty
Cooperative education is not an office down the corridor. The Practice Tutor is faculty, marking workplace learning into the Capability TranscriptCapability Transcript: The verified record of what a graduate can do, fed by evaluators and practice tutors. Select to jump to the glossary. with the same rubric discipline as any course.
Two pathways, one standard
Architects certify Term and Block equivalence course by course, so a choice of pace never becomes a difference of standard.
Assessment anywhere, authority always
With the AI Oral Examination Room, the evaluator role reaches any student on earth, on video, with the same human confirmationHuman authority rule: No AI output becomes a student's result without a named human's confirmation. Select to jump to the glossary. of every result.
Promises with names on them
Every sentence the University says to a student maps to a role, a caseloadCaseload: The number of students, courses or assessments a role is designed to carry, published, not private. Select to jump to the glossary. and a timer, live in Amagezi, not filed in a policy folder.
Grounded inAmagezi platform specification,Assessment and quality framework engines, rubric rules and calibration cadence
Where these claims come from
Sources and grounding.
Nothing on this page is a marketing claim we cannot point at. Each section below names the document, precedent or published commitment it rests on, and where the source is external, you can go and read it yourself.
- S1Internal document
Forward University Faculty Model (founding document)
The Vice-Chancellor's faculty design paper: the nine disaggregated roles, the separation of teaching from evaluation, the human authority rule, and the caseload ratios reproduced on this page verbatim.
What it grounds
- The nine roles and their briefs
- The separation of teaching and evaluation
- Caseload ratios and how each role scales
- Amina's fortnight as an illustration of the mentor role
Internal founding document, quoted on this page and in the downloadable PDF.
- S2External precedent
Western Governors University, founded 1997
The competency-based model established by nineteen United States governors, which separated programme mentors, course instructors, independent evaluators and central curriculum teams and scaled to hundreds of thousands of students.
What it grounds
- The claim that disaggregation is proven at scale, not experimental
- The four results attributed to the precedent
- S3Internal document
Amagezi platform specification
The specification for the assignment, scheduling, promise and audit engines: how each service level becomes a timer, how eligibility for marking is enforced, and what writes to the append-only audit trail.
What it grounds
- Every promise is enforced in software, not policy
- Blind marking and evaluator eligibility checks
- The escalation chain from role holder to Chair to Dean
Internal founding document, quoted on this page and in the downloadable PDF.
- S4Published commitment
Published service commitments
The turnaround targets Forward publishes and reports against, one working day for course answers, 72 hours for marking, a mentor call every two weeks, including where we miss them.
What it grounds
- The service-level table
- The promises table's measurement column
Published by Forward and reported against, including where we miss.
- S5Internal document
Assessment and quality framework
The mastery-grading rules, rubric publication requirements, calibration cadence, and the external examiner's independent sampling remit.
What it grounds
- Rubrics published before a student is marked against them
- Weekly evaluator calibration
- External examiner sits outside the delivery chain
Internal founding document, quoted on this page and in the downloadable PDF.
The hiring structure
The nine roles we recruit against.
Forward does not advertise for generalists. Every academic post is one of these nine, with its own brief, its own caseload and its own accountability.
01
Programme Mentor
Progress, pace and pathway for a named caseload of students, from enrolment to graduation.
02
Course Instructor
Deep subject expertise on one course, taught in both pathways, answering inside one working day.
03
Evaluator
Independent, rubric-bound marking, blind wherever possible, never of a student they taught.
04
Curriculum Architect
Courses as institutional assets: outcomes, rubrics, question banks and AI grounding packs.
05
Practice and Placement Tutor
Sourcing placements, brokering matches, and assessing workplace learning with the employer.
06
AI Teaching Assistant (Musomesa)
Software, supervised: instant grounded help, unlimited practice, and never a mark to give.
07
Programme Chair
One academic owner per programme: competency register, course approvals, progression and award boards.
08
Dean of School
Holds the chairs to account and owns the School's staffing plan across all nine roles.
09
External Examiner and QA Officer
Outside the delivery chain: independent sampling, and audit of separations, turnarounds and calibration.
Why ForwardMeet the mentor model How marking worksPublished commitmentsJoin the team Talk to a person
Share the Faculty Model
A print-ready PDF of this page, the nine roles, the caseload ratios, the service promises, Amina's fortnight and the full FAQ, for applicants, employers and partners.
Download the PDF (8 pages)The words, defined
Glossary.
Search the vocabulary this page uses. Every term is also underlined in the copy above, hover, focus or tap one to read its definition without leaving your place.
Terms are also underlined throughout this page, hover or tap one to see its definition.
10 of 10 terms.
- Disaggregation
- Dividing the traditional professorship into specialised full-time roles.
- Separation of teaching and evaluation
- The rule that no one marks a student they taught, enforced by Amagezi's assignment engine.
- Caseload
- The number of students, courses or assessments a role is designed to carry, published, not private.
- Service promise
- A published turnaround, such as marking in 72 hours, tracked by a timer in Amagezi.
- Blind marking
- Evaluation with the student's identity masked wherever the artefact allows.
- Calibration
- Evaluators marking a shared sample until their scores converge, held weekly.
- Grounding pack
- The architect-published course materials an AI assistant is permitted to answer from.
- Capability Transcript
- The verified record of what a graduate can do, fed by evaluators and practice tutors.
- Competence unit
- The University's measure of learning volume; one unit is fifteen notional hours.
Questions students actually ask
Mentors, evaluators, measurement and the wall.
If a promise cannot survive a direct question, it is not a promise. Here are the direct questions.
Mentors
Evaluators
Measurement
Separation
Still unanswered? Ask us directly , a named person replies.