It is May 2026. A Principal at a Tier 1 engineering college in Tamil Nadu is sitting with the institution’s CIO and IQAC Director. Three things are on the table at the same meeting. The LMS contract is up for renewal in March. The NBA Tier-1 visit is scheduled for September. And a circular has arrived from UGC asking the institution to confirm Academic Bank of Credits readiness for the entering 2027 cohort.
Each of these has its own deadline and its own document trail. None of the three can be fully addressed without addressing the other two. The LMS renewal cannot be decided without knowing whether the platform can issue and accept ABC-compatible credits. The NBA submission cannot be assembled in time without continuous CO-PO attainment data, which the current LMS cannot produce. The ABC readiness cannot be confirmed without protocol-level interoperability the institution has not previously needed.
The Principal asks the CIO a question that is going to be asked in some form at every Indian engineering college over the next two years. What software does this institution actually need to operate under NEP 2020?
This piece is the long answer to that question. The short answer is Intelligent Learning Infrastructure — a category of platform purpose-built for the academic regime NEP creates, and the operating system that the existing LMS estate cannot become through upgrades alone.
The shape of the policy that landed
NEP 2020 reorganised Indian higher education around a small number of structural commitments. Multi-disciplinary mobility between fields and institutions. Multiple entry and exit points within a degree. A national-scale credit economy in which credits earned in one place can be redeemed in another. Vocational and entrepreneurial activity recognised as credit-bearing alongside classroom coursework. Continuous assessment in place of cyclical examinations. Outcome attainment evidence aligned to Washington Accord standards through NBA, expanded continuously through NAAC.
By 2026, the architecture is operational. The Academic Bank of Credits sits inside DigiLocker and accepts entries from registered institutions. The National Higher Education Qualifications Framework has been notified. UGC has issued enabling regulations on dual degrees, online learning, and transnational education. AICTE has rewritten its model curriculum. The Ministry of Education has placed DIKSHA and SWAYAM into the policy text as national digital infrastructure.
What is missing, in most institutions, is the layer between policy and operation. The software that records the credit, tags it correctly against the National Credit Framework, makes it visible to the student in DigiLocker, transmits it to the receiving institution when the student moves, and feeds the institution’s own dashboards for accreditation and management. That layer is supposed to be the LMS. In most Indian engineering colleges, the LMS in place was not designed to do this work.
Why the current LMS estate cannot bridge the gap
A typical engineering college in India today is running an LMS procured between 2014 and 2020. The product was built for content delivery — uploading lectures, holding assignments, recording quiz scores, tracking attendance. In its original architecture, the LMS was a content website with a gradebook attached.
Three architectural assumptions inside that design create the problem with NEP.
- The LMS was built around the single-institution course catalogue. A student belongs to one programme at one college, takes courses defined in that programme, accumulates marks tied to those courses, and graduates with a transcript that lives inside the same system that recorded the marks. There is no provision for a credit earned elsewhere to enter the system as native data, or for a credit earned in this system to leave it as native data.
- The LMS was built around a fixed-duration linear programme. The student enters in year one, progresses sequentially, and graduates in year four. There is no provision for a student to exit after year one with a certificate, return after three years, complete the next year, exit again with a diploma, and complete the degree five years later. Multi-entry-multi-exit is treated as an exception requiring manual intervention rather than as a normal academic journey.
- The LMS was built around content as the unit of academic activity. A lecture is uploaded. A student watches it. A quiz follows. The model handles structured classroom activity well. It does not have a native data model for an incubation cohort, a venture-studio milestone, an industry-issued micro-credential, a vocational module, or an apprenticeship — all of which NEP makes credit-bearing alongside coursework.
Beyond these three architectural assumptions, the foreign-hosted majority of the LMS market faces a separate question. The Digital Personal Data Protection Act of 2023 restricts cross-border transfer of personal data of Indian residents and gives the government discretion over which jurisdictions the data may leave for. Every institution running a foreign-hosted LMS now carries a compliance question that did not exist when the platform was procured. For centrally and state-funded institutions, sovereign deployment has become a question worth asking in accreditation conversations. For private engineering colleges, the question becomes harder when the institution’s research activity touches sensitive datasets.
This is not an argument that foreign LMS platforms are bad. Many of them are competent products that served Indian institutions well in the regime they were procured for. It is an argument that the regime has changed and the products have not.
What UNESCO is independently saying about this gap
The UNESCO Higher Education Global Trends Report 2026, released in May, is the first UN flagship report on global higher education. Its findings make the case for an architectural shift without anyone having to argue it.
UNESCO treats the LMS as legacy infrastructure. Useful for content access. Standard component of institutional IT. Not where the consequential work happens. The consequential work, in UNESCO’s framing, sits in higher education management information systems (HEMIS) and in the integration of learning data with analytics, accreditation, and decision-making.
The data systems are not in place. UNESCO finds that even in OECD countries, less than half of higher education data systems are interoperable at either the institutional or system level. The Indian estate is in a more difficult position because the systems were never procured with interoperability as a requirement.
AI inside higher education is mostly unmanaged. More than 60% of higher education faculty globally already use AI in their work. Eighty per cent say institutional AI guidance is inadequate. Less than 10% of universities had a formal generative AI policy in 2023; only 19% had one by 2025, with another 42% still drafting. UNESCO’s read is direct — AI in higher education is inevitable, mostly unmanaged, and mostly potential rather than practice.
Faculty workload and faculty support is a global blind spot. Fewer than 15% of countries prioritise faculty wellbeing in their national higher education plans. The faculty using AI without institutional guidance are the same faculty being asked to assemble accreditation evidence, run continuous assessment, mentor incubation cohorts, and absorb the operational implications of policy reform.
UNESCO’s closing recommendation is to invest in higher education management information systems, with clearer institutional responsibilities and stronger analytical capacity at both institutional and system levels. For an Indian engineering college reading the report, the practical translation is straightforward. The institutional layer needs to become the system that holds learning, accreditation, faculty, and outcome data together. That is Intelligent Learning Infrastructure.
What an institutional operating system for NEP has to do
NEP 2020 effectively asks for an institutional operating system rather than a content delivery website. Seven structural capabilities are non-negotiable.
Credit interoperability at the protocol layer. The platform must speak LTI — Learning Tools Interoperability — maintained by the international 1EdTech consortium. An LTI-compliant platform can issue credits to and accept credits from other compliant platforms with grades and identity flowing automatically. Without this, ABC integration becomes a custom engineering project at every institution. With it, credit flow is a standard transaction.
Native handling of multi-entry-multi-exit. A student who exits after year one with a certificate, returns three years later, completes year two, exits with a diploma, and returns five years later to complete the degree must appear as a single coherent academic record. Not three separate students. Not a record requiring manual reconstruction. A single record, with the gaps explained in the data rather than papered over.
A credit-bearing object model that includes vocational and venture activity. An incubation cohort milestone, a venture-studio sprint outcome, an industry-issued micro-credential, a vocational module completion, and a classroom course should all be representable as varieties of the same credit-bearing object. The platform’s underlying schema should make this routine, not a workaround.
Continuous outcome attainment. CO-PO attainment data must accumulate as the semester runs, mapped at the lesson-plan stage, tagged to assessments at creation. NBA Tier-1 alignment to Washington Accord requires this. NAAC reaccreditation increasingly assumes it. AICTE’s outcome-based curriculum builds on it. The platform produces this data as a by-product of teaching, not as a project to be staffed up six months before each visit.
Indian data residency by default. All personal data of Indian residents must remain within Indian jurisdiction under DPDP. The platform must run on Indian infrastructure as a default, with the deployment architecture flexible enough to support sovereign cloud for state institutions, multi-tenant SaaS for private colleges, and air-gapped on-premise for institutions with sensitive research portfolios.
An institutional decision layer. Principals, Deans, HODs, IQAC Directors, and Placement Officers need a live view of the institution at the granularity they each operate. This is not a “reports” feature. This is the surface where institutional decisions are made and where accreditation evidence is interrogated. The COEPE — Centre of Excellence for Personalised Education — model is one operational form of this. The point is that the institutional view is a first-class part of the architecture, not a dashboard bolted on top.
An AI capability layer governed by the institution. Adaptive learning paths, retrieval-augmented tutors trained on the institution’s own curriculum, agentic evaluation with human override, dropout prevention through learning analytics. None of this is exotic in 2026. The architectural requirement is that it sits inside the same platform that holds the credit record, under Indian residency, governed by the institution rather than by a foreign vendor’s product roadmap.
This list of seven is what Intelligent Learning Infrastructure is. It is not an LMS feature upgrade. It is the operating system that the regime NEP creates actually needs.
How ILI reshapes the five academic workflows NEP changes
Inside an Indian engineering college, five workflows do most of the consequential work, and NEP reshapes each of them.
The accreditation workflow. The NBA Tier-1 visit, the NAAC reaccreditation cycle, the AQAR submission. In a college without ILI, this workflow is a six-month sprint reconstructing CO-PO attainment from spreadsheets, faculty memory, and best estimates. In a college with ILI, it is a two-week review of documentation that built itself as the semesters ran. UNESCO’s finding that QA frameworks globally struggle to keep up with online learning and micro-credentials applies most acutely to institutions implementing NEP. ILI is the layer that closes the gap.
The outcome attainment workflow. Mapping course outcomes to program outcomes to graduate attributes, calculating direct and indirect assessment, and demonstrating Washington Accord-aligned OBE. Most institutions do this in Excel. The IQAC coordinator inherits formulas from the previous IQAC coordinator. In an ILI, mapping happens at the lesson-plan stage through the TEATAR teaching workflow — Teach, Engage, Assess, Track, Analyse, Remediate. Attainment is calculated continuously. The IQAC coordinator stops being a part-time programmer.
The teaching-to-learning workflow. Plan, teach, assess, see results, decide what to do next. In most colleges this loop runs at the speed of the mid-semester examination — twice a semester. In an ILI, it runs at the speed of the unit. The 7AI Learning Model maps every student across aptitude, attention, learning speed, topic mastery, coding skills, personality, and career fitment. Faculty see assessment patterns within days. Personalised intervention becomes routine rather than aspirational. UNESCO names personalised learning and learning analytics as emerging AI applications globally. ILI is what makes them operational in an Indian college.
The employability workflow. NEP places employability at the centre of graduate quality, and NBA Tier-1 assessment is moving towards employability evidence rather than placement counts. In an ILI, readiness tracking starts in year one and runs across all four years, then continues into alumni feedback at six months, one year, and two years. The placement cell stops being a year-four scramble. The institution can answer specific cohort-readiness questions with data rather than confidence.
The credit interoperability workflow. This workflow exists only because of NEP. A student earns credits at multiple institutions, exits and re-enters, accumulates non-classroom credit-bearing activity, and eventually redeems credits for a degree through the Academic Bank of Credits. An LTI-compliant ILI handles this as a standard transaction. The existing LMS estate, designed for single-institution catalogues, handles it as an exception requiring engineering work. For any engineering college operating in the NEP regime, this workflow alone justifies infrastructure change.
Three forces converging on the institutional decision
The institutional decision is being forced by three things landing in the same eighteen-month window for most colleges.
Software replacement cycles in higher education run on five-to-seven-year contracts. The wave of LMS installations from the mid-2010s is now mechanically due for renewal. A procurement committee deciding whether to extend the same architecture for another five years has to defend that decision against the policy regime that emerged in the meantime.
The DPDP Act has changed the deployment economics of foreign-hosted platforms. Sovereign deployment is now a question worth asking in accreditation conversations, particularly for institutions whose funding model or research portfolio makes data residency consequential. The directional preference for indigenous infrastructure is articulable. It is not yet a hard mandate.
Convergence is changing what the platform has to do. The NEP regime treats curriculum, lab work, certification, mentorship, incubation, and venture activity as a single academic continuum. There is no architectural reason for these to sit in five different vendor systems with five different data-residency profiles. They sit there because they were procured separately by different parts of the institution, before any policy asked them to be the same thing. NEP is now that policy.
Three actions for the institution this year
For a Principal, CIO, or IQAC Director reading this piece, three actions worth taking before the next contract renewal or accreditation cycle.
Audit the current software estate. List every platform currently holding academic, accreditation, vocational, or employability data. For each, name the hosting jurisdiction, the DPDP compliance status, the LTI support status, and the contract renewal date. Most institutions discover that two or three of their critical systems are simultaneously up for renewal, non-compliant, and architecturally unable to support the credit flows NEP requires.
Map the five workflows against the existing stack. For accreditation, outcome attainment, teaching-to-learning, employability, and credit interoperability, identify which system holds the data, who does the integration work between systems, and how long it takes to answer a typical regulator question. The workflows NEP makes mandatory are usually the workflows the current stack handles worst.
Walk through what an ILI deployment would look like at the institution’s specific stage. Edwisely’s ILI is currently deployed under white-labelled platforms at engineering colleges across India including Sreenidhi.Ai (Sreenidhi Institute of Science and Technology), RMK Nextgen Student (RMK Group), SAIL Student, Alliance.Ai, ANITS.Ai, and Vedic.Ai. Each institution operates with the same ILI architecture mapped to its own programmes, NBA tier, syllabi, and stage of NEP implementation.
The Principal who started this piece — the one with the LMS renewal in March, the NBA visit in September, and the ABC circular on the desk — does not have a software problem in the traditional sense. The institution has a category problem. It is being asked to operate inside a regime its current platform was not designed for, and the platform it needs is a different category of software rather than an upgraded version of the one it has.
That category is Intelligent Learning Infrastructure. The institutions that recognise this first will set the template the rest of the system follows.
To see what ILI looks like operationally at a specific Indian engineering college, read the SASTRA University case study [link when live]. To map your institution’s workflows to ILI, request a walkthrough with the Edwisely team [demo link].
Sources:
UNESCO Higher Education Global Trends Report 2026 (CC BY-SA 3.0).
Policy references: NEP 2020, Digital Personal Data Protection Act 2023, UGC and AICTE regulations as notified through 2026, NBA Tier-1 manual, NAAC Revised Assessment and Accreditation Framework
