The SAP ERP is confirmed in the FY25 auditor's report: an integrated system connecting head office, all 10 plants, and other locations, with audit-trail and edit-log functionality. ISO 50001 certification at all 10 plants is confirmed in the ESG Databook FY25, including fortnightly energy performance monitoring against targets, with a current intensity of 8.89 GJ/MT.
IATF 16949 and ISO 9001 at all 10 plants, with AS9100D at Arakonam and Medak, are confirmed in BRSR FY25 — external validation by TUV-Nord and BVQI. The CCTV disclosure is verbatim in BRSR FY25, Principle 3: "CCTV is effectively utilized across all facilities to identify unsafe acts and enable timely behavioural interventions."
The ₹292 Cr INROAD co-investment is confirmed through ATMA programme releases. The iSPEED extension — ₹145 Cr, more than two lakh smallholder farmers — was launched April 2025. Not a single item in this list is flattery. Each is a confirmed fact.
MRF invested where it knew how to create competitive advantage — product intelligence, material science, compound design, the Trichy R&D mini-factory confirmed by MD Rahul Mammen Mappillai at the Global Investors Meet 2024. The ₹225.5 Cr reflects world-class product capability. That is where MRF's competitive advantage lives.
The ₹18.38 Cr in external IT and digital services is not evidence of neglect. MRF runs lean — SAP ERP, quality systems, and plant-floor controls maintained as internal infrastructure, not vendor-managed sprawl. At ₹28,000 Cr revenue across 10 plants, that is a disciplined operating choice.
What the ratio reveals is a specific gap: the operational intelligence infrastructure — the layer that governs what all of that production produces, evidences every decision, and makes the record of MRF's excellence as strong as the excellence itself — has not received the same investment. SkyEdgeAI's non-intrusive OAL requires minimal new IT services expenditure to close that gap. It reads from what already exists.
MRF has fulfilled the EPR obligation. BRSR FY25, Section C, Principle 2 confirms: certificates procured from registered recyclers, annual return filed on the CPCB portal, obligations met under both End-of-Life Tyre and Plastic EPR regulations. Everything done correctly.
CPCB's 100% Y-2 obligation structure — active from FY 2024-25 — means the financial exposure grows proportionally with production volumes. Sivaganga adds a new plant. As production scales, so does the obligation. If CPCB's enforcement desk asked to see the production and import data underpinning the obligation calculation — reconciled against the certificate balance, with recycler compliance verified for each counterparty — how long would that take to produce?
The recycler risk dimension is relevant: investigations by the Punjab PCB found two-thirds of tyre-fuel factories missing required environmental controls. MRF does not control its recycler counterparties' compliance quality. A certificate purchased from a registered recycler with a compliance lapse has a different risk profile than it appears. Year-end manual reconciliation cannot surface this in real time. Slide 10 shows exactly how SkyEdgeAI addresses this.
MRF has committed to carbon neutrality by FY2050, with a medium-term target of 25% Scope 1&2 emissions intensity reduction by FY2028 from the FY2023 baseline. FY25 intensity reached 1.10 tCO2e/MT — improved from 1.23 tCO2e/MT. The trajectory is right. The board-level commitment is real and quarterly-reviewed.
MRF's ISO 50001 energy monitoring data exists. The WHR boiler operates at one plant. The PNG gas steam system operates at another. Nitrogen-based curing is deployed at select plants. The instruments are producing data. But the pathway from those plant-floor instruments to the BRSR Core KPI currently runs through manual compilation assembled at year-end.
'Reasonable assurance' — what SGS India provides — is designed precisely to probe that pathway. When the assurance engagement asks: which instrument produced this number, which calculation logic was applied, which human reviewed and approved it before it appeared in the disclosure — a year-end compilation exercise cannot answer that question with the rigour that FY2028's scrutiny will require. Slide 11 shows the complete, continuously-maintained chain.
EUDR — EU Deforestation Regulation 2023/1115 — requires that before any natural-rubber tyre enters the EU, the importer must demonstrate that the rubber is deforestation-free, legally sourced, and origin-traceable to a specific geolocation. The regulation is current law. MRF exports to 60+ countries; FY25 export revenue reached ₹2,307 crore — 23% growth. EUDR compliance protects every rupee of that growth.
What does not exist — for MRF or any Indian tyre manufacturer currently — is a digital thread connecting plantation geolocation data through the supply chain to the compound batch to the finished tyre to the EU shipment. Without that thread, the EUDR due-diligence statement is a manually assembled declaration rather than a continuously maintained evidence chain.
The same genealogy gap affects OEM qualification and recall readiness. IATF 16949 requires traceability. AS9100D at Arakonam and Medak requires it at aerospace grade, serving Indian Air Force and aviation OEMs. The digital depth of that traceability across 10 plants — how far it reaches, how fast it responds — is the discovery question. Slide 12 shows how SkyEdgeAI builds the complete chain from plantation to customer.
India's most advanced tyre peer received WEF Global Lighthouse recognition — the first tyre factory globally. The outcomes: 54% dispatch turnaround improvement, 25% labour productivity, 18% cycle-time reduction, 31% operating-cost improvement, 47% Scope 1&2 emissions reduction. Achieved through approximately 30 digital use cases deployed in roughly 18 months via Siemens Opcenter.
A second India peer deployed a GenAI manufacturing reasoner over curing machine data — achieving 9% productivity increase on primary equipment, 3% energy reduction — using AWS Bedrock and Amazon Neptune for tyre genealogy. India's most advanced digital tyre operations have moved. The benchmarks are real and recent.
What no company in the benchmark has publicly addressed is the governance layer: model versioning, approval, rollback, drift monitoring. Who authorised the AI recommendation. What evidence supported that authorisation. Whether the resulting decision is tamper-evidently recorded and retrievable. This is precisely the layer SkyEdgeAI builds.
When AI is generating the recommendations — when a digital twin is guiding press scheduling, when computer vision is flagging defects, when an AI reasoner is suggesting curing parameter changes — a new institutional question emerges: who governed the AI? Who authorised the recommendation? What evidence supported that authorisation? Is the record tamper-evident and retrievable?
This question is not answerable by the existing systems. SAP was built for transactions. SCADA was built for control. ISO 50001 monitoring was built for energy performance. IATF quality records were built for quality management. None was designed to govern what AI produces.
SkyEdgeAI's Operational Admissibility Layer is precisely this: a governed intelligence layer built above existing operational systems — reading from them, not modifying them — that captures every decision as it is made, with its evidence, its human authorisation, and its timestamp, in a form that is tamper-evident and continuously retrievable. It does not replace any existing system. It governs what those systems produce.
"Reads" is not "connects" or "integrates." It means read-only. SkyConnect™ does not write to SAP, does not modify SCADA configurations, does not alter IATF quality records. The 10 plants continue operating exactly as they do today. No cutover. No parallel running. No ERP migration. No change request to any existing system.
"Every decision" is precise — not "strategic decisions." Every AI recommendation through the Command & Control Layer. Every quality hold decision in GuardianLedger™. Every curing parameter change approved by a plant engineer. Every EPR certificate acceptance. Every BRSR energy metric reviewed before disclosure. Each one governed and evidenced.
"Governed, evidenced, and defensible" is an operating state — not the outcome of a specific project. The evidence is continuous. It does not require assembly. When a regulator asks, an OEM audits, a board enquires, or SGS India probes — the answer is a retrieval, not a reconstruction.
The ESG & Compliance Layer reads production and import volumes from SAP in real time. The EPR obligation threshold is calculated automatically against actual output — no year-end calculation under deadline, no risk of arithmetic error. The obligation is tracked daily.
Certificate tracking is automated: validity dates, coverage against the obligation balance, transfer status, and recycler due-diligence trail are all maintained as a live record. When a certificate is accepted, it is assessed against the recycler's compliance history — not taken at face value. Counterparties with compliance lapses are flagged before acceptance, not after the annual return is filed.
GuardianLedger™ captures every EPR decision — every obligation match, every certificate acceptance, every annual return data point — as a tamper-evident, timestamped, human-authorised record. The CPCB portal annual return is not assembled from scattered records. It is retrieved. For MRF's multi-plant estate, each of the 10 plants is tracked individually with central oversight. Sivaganga is added without re-architecture.
MRF currently has three confirmed, interacting energy variables: WHR boiler steam (one plant), PNG-fuelled steam (one plant), and nitrogen-based curing (select plants). Currently these are monitored separately. ISO 50001 tracks each individually. None is modelled against the others. The optimisation that comes from understanding how press scheduling affects WHR boiler load, how nitrogen curing assignments affect steam demand, how PNG combustion efficiency responds to the curing queue — none of that exists today.
TwinCore™ models all three together as a unified system. Every optimisation recommendation is human-authorised before any parameter changes. Every approved change is GuardianLedger™-recorded with the operator's decision, the evidence basis, and the timestamp.
The Scope 1&2 carbon accounting layer reads from those instruments automatically — per SKU, per plant, per press, per shift. The current FY25 intensity of 1.10 tCO2e/MT improves toward the 25% reduction target with full visibility at granularities that aggregate-level monitoring cannot provide. When SGS India traces an energy intensity figure for FY2028 assurance, they are reading a continuously maintained chain — not a year-end compilation.
DataGuardian™ constructs the genealogy graph by reading from existing MRF systems — ERP, LIMS, quality records, supplier data — without modifying any of them. The chain from compound batch to dispatch lot is built from what already exists. The EUDR extension connects INROAD's plantation geolocation data to the compound batch assignment — completing the thread that MRF's ₹292 crore upstream investment makes possible but does not, by itself, create.
What a completed genealogy chain changes: a recall notice that currently requires days of manual tracing becomes a query on a graph database — answered before the call ends. An OEM PPAP or AS9100D audit evidence request that currently requires assembling records from LIMS, ERP, quality logs, and press records becomes a GuardianLedger™ retrieval. A EUDR due-diligence statement becomes a continuously maintained chain with verifiable geolocation links.
The read-only integration principle applies throughout. SkyConnect™ reads from every existing system. Nothing is modified. The genealogy chain is built above the existing infrastructure, not inside it. The 10 plants continue operating exactly as today.
The IEC 62443 OT security architecture is overlaid above the existing plant-floor control layer — not inserted into it. Network segmentation, access control, and remote-access governance are implemented as an additional layer. No existing PLC configuration is modified. No SCADA parameter is changed. No historian data is altered.
The ₹18.38 Cr IT services spend context is relevant: an integration approach that requires MRF's IT team to manage a complex new system, maintain new vendor relationships, and support new internal user bases would not fit MRF's operating model. SkyEdgeAI's OAL is designed to be thin — reading from what already exists, adding governance above it, requiring minimal new IT overhead to operate.
For the CIO: the existing ERP architecture, workflows, and permissions are entirely unchanged. The existing historian, SCADA, and quality system configurations are entirely unchanged. The IT team is not acquiring a new system to maintain. They are adding a governance overlay that reads from systems they already manage — with IEC 62443-compliant OT security added as part of the integration architecture.
The WHR boiler efficiency depends on press scheduling: how many presses are opening simultaneously, what condensate is available, what de-superheating load is required. The PNG combustion efficiency responds to the curing queue and to ambient conditions. The nitrogen curing press assignments affect thermal demand per tyre. All three interact. None is currently modelled against the others. ISO 50001 monitors each separately.
India's most advanced tyre peer achieved 9% productivity increase on primary equipment and 3% energy reduction by deploying a GenAI manufacturing reasoner over curing machine data. That reasoner generates insights. What it does not produce is a governed output: the record of who authorised the recommendation, what evidence supported it, what changed as a result. SkyEdgeAI's TwinCore™ is the insight and the governance in the same system.
The bladder RUL prediction component compounds the value: unplanned bladder failures at the curing press create schedule disruption, quality incidents, and maintenance cost simultaneously. Predicting remaining useful life from press cycle data and defect linkage history — and scheduling replacement before failure — addresses three consequences of a single failure mode. Every replacement recommendation is human-authorised and GuardianLedger™-recorded.
GuardianLedger™ answers the EPR evidence question, the BRSR source-chain question, and the OEM audit trail question simultaneously. DataGuardian™ and SkyConnect™ answer the genealogy gap and EUDR thread. TwinCore™ answers the three-variable curing optimisation problem. ESG & Compliance Layer answers all three obligations — EPR, BRSR, and EUDR.
Command & Control Layer answers the governance-of-AI question the benchmark slide raises. InfraOps addresses the OT cybersecurity posture implied by the confirmed PLC/SCADA infrastructure: IEC 62443-compliant architecture, OT asset inventory, AI model versioning, drift monitoring, and rollback.
EdgeVision™ builds on the confirmed CCTV infrastructure across all facilities. The hardware exists, the network exists. AI quality inspection with governed disposition — confidence-scored, human-authorised, IATF-admissible disposition records — adds capability to infrastructure MRF already owns. The GenAI Manufacturing Copilot addresses the concentration of tacit process knowledge in an experienced workforce — capturing shift logs, maintenance history, and troubleshooting playbooks as a governed, queryable knowledge base.
Energy: MRF's confirmed intensity is 8.89 GJ/MT with ISO 50001 at all 10 plants. Energy costs represent approximately 15–18% of tyre manufacturing cost of goods — at ₹28,000 Cr revenue that implies a ₹4,000–5,000 Cr annual energy cost base. India's most advanced peer achieved 47% Scope 1&2 reduction. A conservative 5% improvement is ₹200–250 Cr annually.
Scrap: tyre manufacturing scrap typically runs 2–5% of production value without AI-level genealogy and RCA. At ₹28,000 Cr, each 0.1% of scrap reduction is approximately ₹28 Cr. The three primary scrap drivers — compound batch variation, curing process drift, and inspection inconsistency — are addressed by DataGuardian™, TwinCore™, and EdgeVision™ respectively.
Compliance: the ₹2,307 Cr is MRF's confirmed FY25 export revenue. EUDR is current law. If a major EU customer requires EUDR due-diligence evidence that MRF cannot provide, that customer relationship is at risk. Sivaganga: the 4–8× cost multiplier for governance retrofit vs design-in is an industry-observed ratio. At ₹5,300 Cr over 12 years, architecture decisions made in the next 6–12 months determine the cost and capability profile of the entire programme.
WEF Global Lighthouse recognition for an India tyre manufacturer — the first globally — validates that these outcomes are achievable in an Indian manufacturing context. The 47% Scope 1&2 emissions reduction is the most directly relevant benchmark for MRF's FY2028 25% intensity target. MRF's target requires a smaller proportional improvement than what the peer achieved — and MRF's energy infrastructure provides a more instrumented starting point.
The GenAI manufacturing peer's 9% productivity gain is the curing intelligence benchmark. SkyEdgeAI's TwinCore™ plus DataGuardian™ is architecturally analogous — with the governance layer that peer's solution lacks.
"MRF's foundation is stronger" is not flattery. It is a specific architectural observation. ISO 50001 certification at all 10 plants means the energy monitoring infrastructure is already in place. IATF 16949 and AS9100D means the quality management infrastructure is already in place. SAP ERP means the enterprise data backbone is already in place. The peer that achieved the WEF Lighthouse results started from a less instrumented position. MRF's foundation compresses the time to equivalent outcomes.
The MoU for Sivaganga was signed on 4 March 2026 — confirmed by stock exchange filing. ₹5,300 crore over 12 years. SIPCOT Industrial Park, Sivaganga District, Tamil Nadu — MRF's fourth Tamil Nadu plant. Subject to customised incentive package and statutory approvals.
The IEC 62443 OT security architecture is approximately 4–8× cheaper to implement as a design-in than as a retrofit on a live plant. The GuardianLedger™ evidence chain, starting from first production, creates a plant-of-record from Day 1 rather than requiring retroactive evidence construction. Digital commissioning acceptance criteria — genealogy trace, OEE capture, energy metering, recipe enforcement, cyber segmentation — defined before any equipment arrives means these are go-live requirements, not post-go-live aspirations.
The Sivaganga-as-template argument is equally important. MRF currently operates 10 plants — each likely with different PLC generations, SCADA vendors, historian configurations, and data granularity. Cross-plant benchmarking is not currently possible without data normalisation. Sivaganga, designed with a standard data backbone, becomes the architecture template for that normalisation — a programme that currently has no starting point gets one.
No production system is modified during Phase 0. No plant-floor configuration changes. No ERP change request. No cutover. SkyConnect™ reads; it does not write. The 2–3 pilot plants continue operating exactly as today throughout the 90 days.
What MRF receives at the end of 90 days that it does not have today: a continuously-operating EPR evidence chain from production data to certificate balance; an automated BRSR energy chapter for the pilot plants with source-system lineage; a genealogy baseline for one product family that demonstrates lot isolation capability in minutes; and a curing press digital twin calibrated against actual MRF steam and OEE data — not a template, MRF's data.
The Phase 1 roadmap is agreed jointly at the end of Phase 0, based on confirmed evidence. Not asserted by SkyEdgeAI based on a pre-sale projection. The ROI estimates in Act Three are conservative by design — the Phase 0 baseline replaces estimates with confirmed measurements at MRF's actual scale, in MRF's actual plants, against MRF's actual data quality.
SkyEdgeAI was founded in 2024 with a specific thesis: that the industrial AI market would reach a point — faster than most expected — where capability without governance would become a liability rather than an advantage. That point is now. The regulatory environment (EUDR, BRSR mandatory assurance, CPCB EPR enforcement, OEM audit pressure) has arrived simultaneously across multiple industries.
The InfAIra technology platform is proprietary. GuardianLedger™'s tamper-evident record architecture requires that the evidence chain is not vulnerable to model replacement, API deprecation, or third-party platform decisions. A governance layer built on an open-source LLM or a third-party API is not genuinely tamper-evident. SkyEdgeAI's proprietary architecture ensures the governance integrity of every record it produces.
The founding team brings specific domain depth: Devi Prasad Vuriti's four decades in process-intensive manufacturing (cement, power, fertiliser) provides the OT integration credibility that IT-only governance vendors lack. Kameswara Rao Tangudu's integration architecture experience underpins SkyConnect™. The tyre manufacturing domain mapping was built from scratch — not adapted from a pharma or FMCG template.