It is 11.40 pm and the IQAC coordinator is still on the “SSR 2026” WhatsApp group.
Forty departments. One Excel template. Forty different ideas of what “result declaration date” means.
Someone has posted a screenshot of a result notice instead of the file. Someone else has typed CO attainment figures from memory.
The peer team visit is in eleven weeks, and half of the NAAC accreditation evidence is still a promise.
Every university that has been through a Self Study Report knows this scene. It does not have to repeat.
❗ Why the old rhythm no longer works. In February 2025 NAAC announced its biggest change since 2007: Binary Accreditation followed by optional Maturity-Based Graded Levels, MBGL, from Level 1 to Level 5.
Levels 1 and 2 are assessed largely on digital data and documents. Roughly 90% of accredited institutions still hold CGPA-era grades that expire across 2026-27, and the binary portal has not been fully live on schedule.
Nobody can plan an 18-month sprint around a date that keeps moving. The evidence has to be ready all the time.
The problem is not that universities lack evidence.
The examination system, question-paper system, ERP and LMS generate it every semester.
The problem is that the evidence sits in six systems, in formats no assessor will read, and gets rebuilt by hand every cycle.
Meanwhile NBA asks engineering programmes for the same course-outcome and attainment data in its own format.
Accreditation evidence, by the numbers
This article shows how an accreditation assistant agent turns semester data into an accreditation file that stays current.
It covers what the agent collects, how it maps evidence to NAAC criteria and NBA requirements, and how it finds gaps while there is still time to fix them.
Table of Contents
- → Binary Accreditation, MBGL and CO/PO attainment in plain words
- → Why your assessment platform is the richest evidence source
- → Evidence item, where it already lives, what the agent does
- → What the accreditation assistant agent actually does
- → Guardrails, and six weeks to live
- → Built on platforms that already serve 200+ universities
- → One-semester evidence readiness plan for an IQAC
- → FAQs
🧭 Binary Accreditation, MBGL and CO/PO Attainment in Plain Words
Binary Accreditation is the new first stage of NAAC assessment. An institution is either Accredited or Not Accredited.
There is no A++, A or B grade at this stage.
The idea is to bring many more institutions into the accredited pool with a clear, largely data-driven threshold.
Maturity-Based Graded Levels (MBGL) is the optional second stage.
Accredited institutions climb five levels, from basic compliance at Level 1 to institutions of global excellence at Level 5.
Levels 1 and 2 rely mainly on digital verification of data and documents. Levels 3 to 5 add expert validation.
Under both stages the question a peer team asks is the same: show us the record, not the claim.
| What the peer team asks about | ❌ The five-yearly scramble | ✅ Continuous evidence |
|---|---|---|
| How the file gets built | Rebuilt by hand every cycle | Every semester adds to it automatically |
| Where the numbers come from | Typed from memory into a shared template | Computed from the source systems |
| When gaps surface | Eleven weeks before the visit | End of every semester, listed by owner |
| What backs a claim | An assertion nobody can trace | The source document, cited on the figure |
A Course Outcome (CO) is a statement of what a student should be able to do after a course.
A Programme Outcome (PO) is what a graduate of the programme should be able to do.
Attainment is the measured share of students who reached the target for each outcome, calculated from real exam and internal assessment scores.
NBA’s manuals expect programmes to set attainment targets per course and show, year on year, whether students met them.
Our outcome based education guide covers the mechanics, and this piece on CO-PO mapping covers the setup work.
🗺 Why Your Assessment Platform Is the Richest Evidence Source
Ask any IQAC coordinator which criterion took the most chasing, and the answer is usually teaching, learning and evaluation.
That is also the criterion your examination data feeds most directly.
Consider what a modern assessment platform already records without anyone filling a form.
For engineering programmes, NBA asks for course outcomes, their mapping to programme outcomes, attainment levels and the continuous improvement loop.
NBA is explicit that this must be measured, not asserted. Our guide to streamlining NBA accreditation walks through those challenges.
Most institutions treat this data as an exam-office by-product.
It is accreditation evidence sitting in the wrong drawer.
See our NEP 2020 assessment requirements guide for the policy background, and how NAAC software supports teaching, learning and evaluation for the criterion-level view.
📋 Evidence Item, Where It Already Lives, What the Agent Does
The table below is the heart of the approach.
Each row names an item a peer team or NBA evaluator asks for, the system that already holds it, and the work the agent does so nobody rebuilds it by hand.
| Evidence item | Where it already lives | What the agent does |
|---|---|---|
| CO/PO attainment per course and programme | Question-level scores tagged to COs and Bloom levels in the exam and question-paper system | Computes direct attainment against the programme’s target, charts the trend across semesters, cites the mark file and the paper’s CO map |
| Result declaration timeline | Exam schedule versus result publish timestamps in the on-screen marking system | Computes days from last paper to result for every exam, flags outliers, produces the timeliness table |
| Re-evaluation and grievance transparency | Re-evaluation workflow log: request date, examiner, outcome, closure date | Counts requests per exam, computes average turnaround, shows the share where marks changed |
| Question paper quality and blueprint compliance | Blueprint preflight results and Bloom’s distribution per paper | Summarises compliance per programme, shows the Bloom-level spread, lists papers that failed preflight and how they were fixed |
| Examination integrity | Proctoring logs, identity checks, content-hashed question-paper publish records | Drafts an integrity statement with counts of sessions, flags reviewed and violations confirmed, each linked to its log |
| Reforms in evaluation | Configuration history: on-screen marking adoption date, AI-assisted evaluation modes, moderation rules | Drafts the reform narrative with dates and settings taken from the system, not from memory |
| Student performance analytics | Question-level, topic-wise and batch performance reports | Builds the learning-level analysis for slow and advanced learners, with source report IDs |
Every one of these rows depends on the assessment platform recording the right thing at the source.
An AI question paper generator tags every question to a CO and a Bloom level at setting time, and enforces the blueprint before release.
That is what makes attainment computable rather than estimated. Bloom’s taxonomy tagging is the part most institutions bolt on too late.
An on-screen marking system that stores per-question marks with a full audit trail is what makes timeliness and re-evaluation transparency provable.
Onscreen marking already carries weight in the NAAC process; the agent is what turns it into a filed, cited record.
⚙ What the Accreditation Assistant Agent Actually Does
The accreditation assistant in Splashgain’s education agent pack has one job.
NAAC and NBA evidence gathered continuously, not in an 18-month scramble.
It works in four steps, and each step is visible to the IQAC. It is the natural sixth member of the agent set every university is now assembling.
Step three is the one that ends the scramble.
The IQAC gets a short, owner-wise list while there is still a semester left to act on it.
A peer team member asks how many re-evaluation requests you received in 2025-26 and how fast you closed them. The answer comes back with the log attached.
- CO and PO attainment computed from question-level scores.
- Result declaration timelines per exam, computed, not recalled.
- Re-evaluation counts and turnaround from the workflow log.
- Every figure cited to the source document behind it.
🛡 Guardrails, and Six Weeks to Live
An accreditation file is a formal record of the institution.
So the agent carries the same guardrails Splashgain applies to anything touching money, contracts or a customer promise.
Draft-and-approve is the default: the agent proposes, a named IQAC member approves.
Every collection run, mapping change and draft is written to an audit log, so the accreditation committee can show who approved what and when.
Data stays where the university decides.
The agent runs in the institution’s own cloud or VPC, or on Splashgain’s CERT-IN certified stack, and no data trains a public model.
Student-level marks follow the institution’s DPDP retention rules, and reports carry the university’s name and logo, not a vendor badge.
Discover. Two workshops to name the metric, and days to assemble a first criterion pack.
Prioritise. Two or three agents, not twenty. The criterion with the most manual chasing goes first.
Build. Two-week sprints on real data in a sandbox, against your own last two semesters.
Deploy. Monitoring, logging, rollback, and a named owner inside the IQAC.
Operate. Running under SLA, with the metric reported monthly to the IQAC.
Institutions that want proof before commitment can start with the Agent Readiness Sprint: fixed fee, 10 working days, credited against deployment.
It ends with five deliverables: process and data discovery, a prioritised agent roadmap, an ROI model built from your own volumes, an executive readout, and one working prototype on your own data.
For an IQAC the natural prototype is a CO-attainment evidence pack for one programme, built from its last two semesters of results.
🏛 Built on Platforms That Already Serve 200+ Universities
The reason this is not another AI pilot is that the evidence sources already exist at scale.
Eklavvya’s outcome based education support already produces CO/PO attainment, Bloom-level analytics, question-level analytics and batch performance.
The team has provided NAAC accreditation assistance to universities using exactly that data.
When result declaration is a matter of days, the timeliness metric writes itself.
“With Eklavvya’s onscreen marking system, the checking process of the answer sheet got a lot easier and smoother.”
– Prof. Shubhashish Goswami, Controller of Examination, Dev Bhoomi Uttarakhand University, where results moved from 45 days to 8
✅ One-Semester Evidence Readiness Plan for an IQAC
Whether or not you deploy an agent this year, these eight steps move an institution from scramble to readiness in one semester.
They are also the preparation that makes an agent useful on day one.
One-semester evidence readiness checklist
- Name one evidence owner per department and one metric per criterion, agreed in writing by week 2
- Publish CO statements and CO/PO maps for every course in the question-paper system, not in a Word file
- Require every question to carry a CO tag and a Bloom level at paper-setting time, and enforce the blueprint before release
- Store per-question marks for every exam, so attainment is computed from scores rather than estimated from grades
- Record exam-schedule and result-publish timestamps in one system, and review days-to-result after every exam
- Run re-evaluation through a logged workflow with request, examiner, outcome and closure dates
- Fix attainment targets per course before the semester starts, and review shortfalls at the mid-semester IQAC meeting
- Build one criterion pack end to end this semester, from source data to cited draft, and time how long it took
Item eight is the one most institutions skip.
Timing one criterion pack gives the IQAC its baseline metric, and gives any agent engagement a number to beat.
If the exam calendar itself is the bottleneck, start with university examination management and onscreen evaluation before layering an agent on top.
🚀 The Way Forward: Evidence as a Habit, Not an Event
NAAC’s Binary Accreditation and MBGL, and NBA’s insistence on measured attainment, point the same way.
Accreditation is becoming a continuous, data-verified process rather than a once-in-five-years event.
Institutions whose evidence is always current will move up the maturity levels. Institutions that rebuild the file every cycle will spend the same 18 months again.
The good news is that the raw material is already there.
Every semester of exams on a modern assessment platform produces the CO attainment, timeliness, transparency and integrity records a peer team asks for.
An accreditation assistant agent keeps them collected, mapped, gap-checked and drafted, with a human approving every page.
Key Takeaways
- NAAC’s Binary Accreditation and MBGL, with Levels 1-2 assessed largely digitally, reward institutions whose evidence is continuously ready.
- Around 90% of accredited HEIs hold grades expiring across 2026-27, and portal timing has slipped, so waiting for a date is not a plan.
- Your assessment platform already holds evidence for teaching-learning-evaluation, curriculum and governance criteria, and for NBA outcomes and attainment.
- CO/PO attainment is computable when questions are tagged to COs and Bloom levels at setting time and marks are stored per question.
- An accreditation assistant agent collects, maps, gap-finds and drafts; the IQAC approves; nothing is submitted by the agent.
- Every drafted number cites its source document, so a peer-team question is answered with the record attached.
- Guardrails: draft-and-approve, audit log, the university’s own cloud, DPDP-aligned retention, white-labelled reports.
- One agent live in about six weeks, or a 10-day Readiness Sprint to prove it on one programme first.
- One programme’s CO-attainment evidence pack, every number cited.
- An ROI model built from your own exam volumes and staff hours.
- A gap list by criterion, with an owner against each item.
- A prioritised roadmap of which criterion packs to automate next.
❓ Frequently Asked Questions
What is NAAC Binary Accreditation?
Binary Accreditation is NAAC’s revised first-stage outcome. An institution is either Accredited or Not Accredited, instead of receiving a CGPA-based grade such as A++ or B.
It is followed by the optional Maturity-Based Graded Levels, where accredited institutions are placed on Levels 1 to 5 based on evidence of improving quality.
What are Maturity-Based Graded Levels (MBGL)?
MBGL is the second, optional stage of the new NAAC framework. Accredited institutions progress through five levels, from basic compliance at Level 1 to institutions of global excellence at Level 5.
Levels 1 and 2 are assessed largely on digital data and documents, while Levels 3 to 5 add expert validation. Continuous, well-organised evidence is what moves an institution up the levels.
What is a CO PO attainment report?
A Course Outcome (CO) is what a student should be able to do after a course, and a Programme Outcome (PO) is what a graduate of the programme should be able to do.
An attainment report shows, from actual exam and internal assessment scores, what share of students reached the target for each outcome. NBA and NAAC both expect this evidence for outcome based education.
Does the accreditation assistant agent submit anything to NAAC or NBA?
No. The agent collects, maps, gap-finds and drafts, and every number in a draft cites its source document.
The IQAC or accreditation committee reviews and approves each section. Nothing is uploaded to a NAAC or NBA portal by the agent.
It runs on the university’s own cloud, keeps an audit log of every action, and follows the institution’s DPDP retention rules.
How quickly can an accreditation assistant agent be live?
Splashgain’s engagement model puts one agent live in about six weeks: Discover, Prioritise, Build, Deploy, Operate.
Institutions that want proof first can start with a fixed-fee, 10-working-day Agent Readiness Sprint, ending with one working prototype on their own data.
📚 Related Reading
Last updated: 24 August 2026. NAAC and NBA framework details reflect publicly reported developments as of the publish date and should be checked against the latest official manuals before submission.
