Trends & Innovation · July 3, 2026 · 16 min read
AI Personalisation in Corporate Gifting 2026: From Name Drops to Predictive Gifting
How Indian HR, brand and rewards teams are moving beyond name-engraved mugs to AI-driven predictive gifting — recipient-preference models, generative artwork, dynamic kit assembly, DPDP-safe data pipelines, GST/HSN discipline, Section 194R capture and a 90-day rollout playbook for FY 2026.
By Manjitt S Chawla, Co-Founder, Corpokit
The engraved-name mug ran its course sometime around 2019. It still ships in enormous volumes — because it is safe, cheap and photographs well on LinkedIn — but the Indian HR and rewards teams that measure recipient satisfaction have quietly moved on. In 2026, the leading question in every gifting brief we receive from CHROs in Delhi NCR, Mumbai and Bengaluru is not what should we engrave? It is how do we make each recipient feel like this gift was chosen for them?
AI personalisation is the answer, but the term is doing a lot of work. It covers everything from a generative artwork engine that mints a unique illustration per recipient, to a recommendation model that predicts which of 40 curated SKUs a given employee is most likely to keep and use. Some of it is genuinely useful. Some of it is marketing dressed as machine learning. This guide separates the two.
We walk through the four generations of gifting personalisation — from name-drop to predictive — and lay out the FY 2026 procurement standard for Indian corporates: what data you can legally collect under the Digital Personal Data Protection Act 2023 (DPDP), what consent language works, how a redemption-portal and dynamic-kit-assembly workflow actually runs on a warehouse floor, GST 2.0 and HSN discipline per SKU, Section 17(5)(h) ITC reversal posture, Section 194R capture on non-employee cumulative spend, Rule 3(7)(iv) ₹5,000 perquisite discipline on employees, budget bands from ₹450 to ₹4,500 per kit, and the 90-day rollout Corpokit ships pan-India for programmes of 500 to 20,000 recipients. If your FY 2026 rewards or annual-gifting calendar is up for planning, this is the brief.
The Four Generations of Gifting Personalisation — Where Your Programme Sits Today
Every corporate gifting programme in India today sits at one of four personalisation generations. Knowing which generation you are on — and which is worth moving to — is the first strategic decision, not the SKU choice.
Generation 1 — Name Drop. A single SKU across the entire recipient base, personalised only with a laser-engraved or UV-printed name. Diwali mug × 2,000 employees. Cheapest per kit, fastest to produce, lowest logistics complexity. Perceived-value uplift over an un-personalised kit: ~1.2×. Fine for scale kits under ₹350 where every recipient gets the same object. Not fine for premium tiers where the recipient expects the gift to feel considered.
Generation 2 — Segment Personalisation. Two to five kit variants across the recipient base, split by segment — grade (junior / senior / leadership), geography (North / South / East / West), gender (M / F / non-disclosed), festival (Diwali / Eid / Christmas / Pongal). Same core SKU, different assemblies. Perceived-value uplift: ~1.5×. Most Indian corporates with 500–2,000 recipients are at G2 today. Ceiling — recipient still doesn't feel individually considered.
Generation 3 — Preference-based (Choice). Recipient logs into a redemption portal and picks 1 of 4–8 curated kits or SKUs, or 3 of 12 individual items into a build-your-own kit. Recipient exercises control; satisfaction NPS jumps because psychological ownership shifts to the recipient. Perceived-value uplift: ~2.1×. Requires a redemption portal, an eligibility engine (grade × budget × SKU pool), a redemption-window workflow (10–21 days with reminders), and warehouse dynamic-kit-assembly capability.
Generation 4 — Predictive (AI recommends). A recommendation model predicts the top-1 SKU per recipient using consented signals — tenure, role, geography, past-kit engagement (kept / donated / returned), wishlist / lightweight-survey inputs, festival preference. Recipient can accept the recommendation with one tap or over-ride from the pool. Perceived-value uplift: ~2.7–3.1× when the model is well-trained; ~1.8× when the model is under-trained (worse than G3). Requires everything G3 needs, plus a model training pipeline, historical satisfaction data (or a rule-based fallback for cold-start), and a monitoring dashboard for over-ride rates.
How to choose. Programmes under ₹350 per kit — stay at G1 or G2; the personalisation ROI does not justify portal cost. Programmes ₹500–₹1,500 — move to G3 preference-portal; the uplift is real and the platform fee (₹15–₹25 per recipient per year) is <2% of programme cost. Programmes ₹1,500–₹4,500 (long-service, Annual Day winners, executive Diwali) — G4 predictive is the fastest-growing tier in India in FY 2026 because the per-kit budget carries the platform fee comfortably and the satisfaction uplift is measurable in NPS. Note: do NOT jump G1 → G4 in one cycle. Move G1 → G3 first, capture 12 months of satisfaction data, then train the predictive model on it.
DPDP Act 2023 — What Data You Can Collect, What You Cannot
The Digital Personal Data Protection Act 2023 (with Draft Rules 2025 published) is the compliance envelope for every AI-personalised gifting programme in India. The Act applies whenever you process the personal data of any 'Data Principal' in India — including your own employees. Treating employees as a special case is a mistake; consent, purpose limitation and withdrawal rights apply exactly the same way to them.
The six DPDP tests every gifting programme must pass. (1) Informed consent — plain-language, itemised, granular. (2) Specified purpose — 'personalising your annual gifting kit for FY 2026'; not 'improving employee experience'. (3) Data minimisation — collect only what the personalisation actually needs. (4) Purpose limitation — data collected for gifting cannot silently flow into marketing, appraisal or HR analytics. (5) Storage limitation — retain for the gifting cycle plus a defensible audit window (typically 12 months post-dispatch). (6) Withdrawal right — the recipient can withdraw consent at any point and the personalisation reverts to a default kit.
Lawful to collect (with consent). Dietary preference (veg / non-veg / vegan / Jain / allergy). Apparel size. Delivery address (for direct-to-home shipping). Festival preference. Colour / style preference. Redemption-portal selections. Kit-satisfaction feedback (NPS). Past-kit engagement signal (kept / donated / returned — inferred from a lightweight follow-up survey, not surveillance).
Not lawful without a separate purpose-specific consent — and rarely justifiable for gifting. Health data (diabetes / dietary conditions beyond veg-non-veg — collect only if the kit contains health-relevant items). Financial data. Religion, caste or community. Biometric or genetic data. Location tracking beyond delivery address. Family details, dependents or spouse data.
Consent template essentials. Clear purpose statement. Itemised data list. Withdrawal instructions with a working contact. DPO / grievance officer email. Retention schedule. Third-party processor disclosure (e.g. Corpokit as the platform processor). No pre-ticked boxes. No bundled consent (gifting personalisation must not be bundled with marketing consent). Signed dual-column consent record maintained for the retention window.
HRMS integration constraint. Do not pull the entire HRMS record into the personalisation platform. Pull only the fields the model needs — employee ID, grade, tenure bucket (not exact join date), geography (city, not exact address), and consent state. Delivery address collected via portal, not from HRMS, so the recipient controls it.
The Recommendation Model — What Signals Actually Work
Predictive gifting is only as good as the model. Most Indian implementations fail because the model is trained on the wrong signals — usually because the right signals were never captured. Here is the signal hierarchy that actually predicts kit satisfaction in Indian corporates.
Tier-1 signals (highest predictive weight). Prior-kit satisfaction NPS on comparable SKUs. Explicit preference-portal selections from prior cycles. Wishlist entries from a redemption portal. Kit-return / kit-swap history. These signals directly encode recipient preference and beat every demographic proxy.
Tier-2 signals (moderate weight). Tenure bucket (0–2 / 2–5 / 5–10 / 10+ years). Grade / role level. Geography (city or region — matters for temperature-appropriate SKUs like insulated bottles vs cooling towels). Function (engineering / sales / operations — sales teams engage differently with luggage and tech vs stationery).
Tier-3 signals (weak but useful for cold-start). Age bucket (with consent). Festival preference. Language preference (for regional artwork and message). Dietary preference (for hamper personalisation).
Signals to avoid. Gender inference. Marital status. Family composition. Salary or CTC. Performance rating. Any of these entering a gifting recommendation model creates fairness, bias and compliance risk far larger than the incremental predictive lift.
Cold-start (year one, no historical data). Use a rule-based fallback — tenure × grade × geography × dietary preference. Layer in preference-portal selections captured live; move to a lightweight collaborative-filtering model at 6-month mark and a full recommender at 12 months. Do not launch a black-box deep-learning model against a cold-start dataset; it will over-fit noise and the over-ride rate will spike above 15%, which destroys the perceived-value story.
Model governance. Publish an internal model card — what signals it uses, what it does not, what fairness tests it passes (equal recommendation distribution across gender and tenure buckets after controlling for grade), and how the recipient can over-ride. Align with NITI Aayog's Responsible AI for All (RAI) principles. Log every recommendation for a defensible audit trail. Recipient over-ride rate is the single-best model-health metric — target <10% steady-state; above 15% means the model needs retraining or the SKU pool is wrong.
Redemption Portal + Dynamic Kit Assembly — The Operational Backbone
AI personalisation lives or dies at the warehouse, not in the model. Two systems must work in lock-step — the recipient-facing redemption portal, and the dynamic kit assembly line.
The redemption portal. Employee-authenticated (SSO with the corporate identity provider). DPDP consent captured on first login. Recipient sees either (a) 4–8 curated kits/SKUs to pick from (G3), or (b) an AI-recommended top-1 SKU with an 'accept' button and an 'over-ride' link to the full pool (G4). Delivery address is captured or confirmed. Redemption window 10–21 days with day-7 and day-12 reminders. Consent-refusers and no-shows receive a default kit assigned by the eligibility engine. Portal must be mobile-first (>70% Indian corporate recipients redeem on mobile). Must render in Hindi + English at minimum; add regional languages for programmes with a strong regional cohort.
The eligibility engine. Filters the SKU pool per recipient by grade × budget × geography × dietary × size availability. A ₹4,500 executive premium kit should not surface to a junior recipient on a ₹1,200 budget; nor should a beef-jerky hamper surface to a vegetarian dietary preference. The engine also enforces the per-employee cumulative Rule 3(7)(iv) ₹5,000 exemption — if the recipient has already received ₹3,800 of gifts this FY, only sub-₹1,200 kits should be surfaced (or a warning shown to HR).
Dynamic kit assembly at the warehouse. Pick-list export from the portal freezes the recipient × SKU manifest. Dedicated packing bays with barcode scanners; each SKU is scanned into a kit-level barcode, with the WMS blocking packing if a wrong SKU is scanned. Throughput 180–220 kits/day/bay (versus ~600/day for a static kit) — plan bay-count accordingly. Kit-level QC (100% inspection on Gen-4 predictive; AQL 2.5 on Gen-3 preference).
Generative artwork print-on-demand. Per-recipient card, box sleeve or insert generated by a template + variable-field engine (recipient name, message, occasion-specific illustration). Print integrated with the packing bay — the artwork emerges from a digital press adjacent to the packing station and is paired with the kit before sealing. Do NOT batch-print artwork ahead — pairing errors between artwork and kit are the single-most-common Gen-4 dispatch failure. Cost per artwork insert: ₹8–₹35 depending on substrate.
Ship-out and last-mile. Individual recipient-labelled shipments to home address (WFH cohort) or master cartons sub-labelled by branch for on-site distribution. E-invoice IRN per invoice with HSN per SKU (not aggregated). E-way bill per consignment above ₹50,000. Delivery confirmation captured back into the platform for the satisfaction NPS trigger.
Compliance Envelope — GST 2.0, HSN Discipline, Section 194R and Rule 3(7)(iv)
AI personalisation does not create new compliance categories; it makes the existing categories more traceable — for better or worse. Traceability is only useful if the compliance workflow keeps up.
GST 2.0 and HSN per SKU. Under the CBIC GST 2.0 rate rationalisation notifications (effective 22 September 2025), every SKU in a personalised kit carries its own HSN and rate — drinkware 3924/7013 at 18%, apparel 61/62 at 5–12%, tech 8517/8518 at 18%, notebooks 4820 at 12%, F&B per hamper HSN. Do NOT invoice a kit at a single 'gift kit' HSN — this fails the Rule 46 test at CA audit. The personalisation service (engraving, UV print, generative artwork) and platform / recommendation-engine fees fall under SAC 998434 / 998439 at 18%. E-invoice IRN mandatory above threshold; the personalised nature of the invoice line item does not exempt it.
Section 17(5)(h) ITC reversal. ITC on all gifting SKUs distributed free to employees, channel partners and clients is blocked. Book the reversal in the same GSTR-3B period as the dispatch — do not defer. The personalisation platform should push a monthly ITC-reversal report to the finance team.
Section 194R on non-employee cumulative spend. For influencer PR kits, channel partner rewards, dealer / distributor gifting, and freelance-creator kits — where cumulative gift-FMV to a single recipient exceeds ₹20,000 in a FY, deduct 10% TDS under Section 194R and report in Form 26Q. AI personalisation is a gift — the trigger is FMV, not the personalisation cost. Capture recipient PAN at consent-collection stage on the portal; the platform surfaces cumulative recipient-level spend to the finance team in real time. See our Section 194R guide.
Rule 3(7)(iv) ₹5,000 employee exemption. Non-cash gifts up to ₹5,000 per employee per FY (aggregate across all gifting events — joining kit, Diwali, birthday, anniversary, Annual Day) are excluded from perquisite. Above ₹5,000, the FULL value (not the excess) becomes taxable perquisite via Form 16. The personalisation platform must surface per-employee cumulative FMV to payroll on a monthly basis and warn the eligibility engine when a recipient is approaching the ₹5,000 threshold.
Invoice discipline. Vendor invoice must list every SKU per kit at its own HSN and GST rate; personalisation service on a separate SAC line; e-invoice IRN; recipient count with corresponding kit-value bracket for Rule 3(7)(iv) reconciliation. See our invoice compliance CA audit guide.
UK Bribery Act / FCPA overlay — if the recipient cohort includes foreign officials, foreign customers or foreign channel partners, AI personalisation makes the per-recipient tracking easier for anti-bribery due diligence. See our UK Bribery Act & FCPA guide.
The 90-Day Rollout for 2,000 Recipients — Brief, Consent, Portal, Pack, Close
A well-run Gen-3 or Gen-4 programme is a 90-day exercise for a 2,000-recipient cohort. Compressing below 60 days is possible only if HRMS integration and the SKU pool are pre-existing.
Weeks 1–2 — Brief and SKU pool. Occasion, recipient count, segmentation, budget per tier, SKU pool of 12–20 across drinkware, apparel, tech, notebooks, hampers. Personalisation generation decision (G3 preference vs G4 predictive). Generative-artwork brief. Sign-off from HR head, brand and finance.
Weeks 3–4 — DPDP, HRMS, portal, model. DPDP-compliant consent template drafted with legal. HRMS field mapping (employee ID, grade, tenure bucket, geography, consent state — minimum viable fields). Redemption portal branded and staged. Recommendation model trained on prior satisfaction data if available, or rule-based fallback locked.
Weeks 5–6 — Pilot. 100-recipient pilot cohort across two locations. Capture consent rate, portal-completion rate, redemption behaviour, warehouse pack-error rate, satisfaction NPS. Iterate SKU pool and model weights. Full-programme go/no-go sign-off.
Weeks 7–9 — Full consent, redemption. Consent window 10–14 days with two reminders. Redemption window 14 days with day-7 and day-12 reminders. Consent-refusers and no-shows assigned default kits by the eligibility engine. Freeze the recipient × SKU manifest with HR sign-off.
Weeks 10–11 — Dynamic kit assembly. WMS pick-list. Dedicated packing bays at 180–220 kits/day/bay. Barcode-scan enforcement. Generative artwork print-on-demand paired to each kit before sealing. Recipient-labelled shipping or branch-sub-labelled master cartons. E-invoice IRN per invoice with HSN per SKU. E-way bills per consignment above ₹50,000.
Week 12 + 30 days — Closure and audit archive. ITC reversal under Section 17(5)(h) booked in GSTR-3B. Section 194R Form 26Q for non-employee cumulative >₹20,000. Rule 3(7)(iv) per-employee cumulative posted to payroll. Satisfaction NPS survey (target 2.3–3.1× uplift vs prior name-drop baseline). Model retraining dataset export (consent-scoped, purpose-limited). Full audit archive — PO with per-SKU HSN, DPDP consent registry, portal audit log, recipient × SKU manifest, dispatch photographs, e-way bills, ITC reversal journal, Form 26Q, DPDP retention schedule.
Common mistakes — (1) skipping DPDP consent because 'they're our employees'; (2) launching G4 predictive with no historical satisfaction data (over-ride rate spikes past 15%); (3) invoicing personalised kits at a single 'gift kit' HSN — fails at Rule 46 audit; (4) not booking Section 17(5)(h) ITC reversal in the same GSTR-3B period as dispatch; (5) missing per-employee cumulative Rule 3(7)(iv) tracking and mis-reporting perquisite on Form 16; (6) batch-printing generative artwork ahead of packing and creating recipient-artwork pairing errors; (7) collecting HRMS fields the model does not need (violates DPDP data-minimisation); (8) no recipient over-ride path on the portal — turns G4 into a black box and destroys perceived value. Contact Corpokit or call +91 9999012429 / +91 9310384204 to brief your FY 2026 AI-personalised gifting programme.
Frequently Asked Questions
What is AI personalisation in corporate gifting, and how is it different from a name-engraved mug?
AI personalisation is the use of data — recipient preferences, tenure, role, location, past-kit engagement, festival calendar — to decide either (a) which SKU each recipient receives from a curated set, or (b) what artwork/message is generated for their kit, or (c) both. A name-engraved mug is level-1 personalisation: the recipient's name is printed but the SKU is identical for every person. AI personalisation is levels 3–4: the SKU itself, the artwork, the box message and even the packaging vary by recipient. In FY 2026, credible AI personalisation runs on either a preference-selection portal (recipient picks from 4–8 options) or a predictive model (system recommends the top-1 SKU based on signals), plus a generative artwork engine for individual kit inserts, cards and packaging.
What are the four generations of gifting personalisation in India?
Generation 1 — Name Drop: A single SKU, name laser-engraved or UV-printed. Cheapest per kit, lowest perceived-value uplift, most-common in India. Generation 2 — Segment Personalisation: Tier-based variants (grade, geography, gender, festival). Two to five kit variants across the recipient base. Moderate uplift. Generation 3 — Preference-based (choice): Recipient logs into a redemption portal, chooses 1 of 4–8 curated kits or SKUs. High perceived value because recipient exercises control. Requires a portal, an eligibility engine and a redemption-window workflow. Generation 4 — Predictive (AI recommends): An ML recommendation model surfaces the top-1 SKU per recipient using signals (tenure, role, past kit engagement, wishlist entries, DPDP-consented preferences, geography, festival). Recipient can accept or over-ride. Highest perceived value and highest logistics complexity — requires dynamic kit assembly at the warehouse and a return / swap workflow.
What data can we legally collect from employees for AI-personalised gifting under the DPDP Act 2023?
The Digital Personal Data Protection Act 2023 requires (a) informed consent, (b) specified purpose, (c) data minimisation, (d) purpose limitation, (e) storage limitation, and (f) the ability to withdraw consent. For personalised gifting you can lawfully collect — with consent — preference-selection responses, dietary restrictions (veg/non-veg/vegan/allergy), size (for apparel), delivery address, festival preference and past-kit satisfaction. You should NOT collect health data, financial data, religion, caste, or biometric data for gifting purposes. Consent must be plain-language, granular (recipient can consent to gifting personalisation without consenting to marketing), time-bound, and withdrawable. Retention should be limited to the gifting-programme cycle (typically one financial year). A DPO / grievance officer contact must be surfaced. Corpokit's default consent template covers all six DPDP heads and is aligned with the Draft Rules 2025 iteration.
How does dynamic kit assembly work at the warehouse for AI-personalised programmes?
Dynamic kit assembly replaces the traditional 'pack one kit design × N recipients' with 'assemble N unique kits from a pool of M SKUs'. The workflow: (1) recipient dataset with SKU-per-recipient assignment lands in a WMS pick-list; (2) each recipient's kit is built on a dedicated packing bay with a barcoded manifest sheet; (3) picker scans each SKU into the kit barcode — the WMS blocks packing if a wrong SKU is scanned; (4) generative artwork insert (per-recipient card/box sleeve) is printed on-demand and paired with the kit; (5) kit is sealed, labelled with recipient name and shipping address, and staged into the courier manifest. Throughput drops from ~600 kits/day (static) to ~180–220 kits/day (dynamic) per packing bay, so warehouse capacity planning is critical. Rejection rate must be <0.3%; any higher and the model / portal has upstream data issues.
What is the GST, HSN and ITC treatment for AI-personalised gifting kits?
HSN is per-SKU, not per-kit. Under GST 2.0 (effective 22 September 2025), a personalised kit invoice must list each component at its own HSN with its own GST rate — drinkware under 3924/7013, apparel under 61/62, tech accessories under 8517/8518, notebooks under 4820, hampers per F&B HSN. Do NOT aggregate the kit at a single HSN. Personalisation service (engraving, UV print) and platform / recommendation-engine fees fall under SAC 998434 / 998439 at 18% GST. ITC on gifts distributed free to employees, channel partners and clients is blocked under Section 17(5)(h) CGST — reverse in the same GSTR-3B period as dispatch. E-invoice IRN mandatory above turnover threshold. See our invoice compliance for CA audit guide and GST on corporate gifts guide.
Does Section 194R apply on AI-personalised gifts to channel partners, and what about the Rule 3(7)(iv) ₹5,000 employee exemption?
Yes on both. For non-employees (channel partners, dealers, distributors, clients, influencers, freelance creators) — where the cumulative fair-market value of gifts to a single recipient exceeds ₹20,000 in a financial year, deduct 10% TDS under Section 194R and report in Form 26Q. AI personalisation makes recipient-level tracking easier, not harder — the redemption portal captures PAN at consent-collection stage. For employees, non-cash gifts up to ₹5,000 per employee per financial year (aggregate across all gifting events — Diwali, Annual Day, joining kit, birthday, anniversary) are excluded from perquisite under Rule 3(7)(iv). Above ₹5,000, the full value (not the excess) becomes taxable perquisite via Form 16. The personalisation platform should surface per-employee cumulative spend to the HR/payroll team in real time. See our Section 194R guide.
What are the typical price bands and platform fees for AI-personalised gifting programmes?
Kit price bands: entry (preference portal, 4 SKU choices, standard packaging) — ₹450–₹900 per recipient at MOQ 500. Standard (preference portal, 6–8 SKU choices, generative artwork insert, branded box) — ₹1,200–₹2,200 at MOQ 500. Premium (predictive engine with recipient over-ride, 12+ SKU pool, dynamic kit assembly, personalised video card via QR) — ₹2,500–₹4,500 at MOQ 1,000. Platform fee (redemption portal, eligibility engine, WMS integration, DPDP consent capture, recommendation model): ₹15–₹40 per active recipient per year, plus a one-time integration fee of ₹1,50,000–₹6,00,000 depending on HRMS integration depth. Generative artwork cost per recipient: ₹8–₹35 depending on model complexity and print substrate. Return / swap workflow: 4–7% of recipients typically over-ride the prediction; contract this as scope.
What is the 90-day rollout plan for a 2,000-recipient AI-personalised gifting programme?
Weeks 1–2: brief lock — occasion (Diwali / Annual Day / joining kit / long-service), recipient count and segmentation, budget per tier, SKU pool curation (12–20 SKUs across categories), generative-artwork brief. Weeks 3–4: DPDP consent template drafted with legal, HRMS integration for recipient dataset, redemption portal branded and staged, recommendation model trained on historical satisfaction data (if available) or launched with rule-based fallback. Weeks 5–6: soft launch to a 100-recipient pilot cohort, capture consent, redemption behaviour, over-ride rates and satisfaction NPS; iterate model and SKU pool. Weeks 7–9: full-cohort consent capture window (10–14 days), redemption/selection window (14 days), reminders. Weeks 10–11: WMS pick-list generation, dynamic kit assembly at 180–220 kits/day per bay, generative artwork print-on-demand, kit sealing and dispatch with per-consignment e-way bills. Week 12: post-programme reconciliation — ITC reversal booked in GSTR-3B, Section 194R Form 26Q per non-employee recipient, Rule 3(7)(iv) per-employee cumulative posted to payroll, satisfaction NPS survey, model retraining dataset export. Archive the full file for FY-close audit.
Citations
- Digital Personal Data Protection Act 2023 (DPDP)
- Draft DPDP Rules 2025 — Consent, Purpose Limitation, DPO
- CGST Act 2017 — Section 17(5)(h), Section 31, Rule 46
- CBIC GST 2.0 Rate Rationalisation Notifications (22 September 2025)
- Income Tax Act 1961 — Section 17(2), Rule 3(7)(iv), Section 194R
- CBDT Circular 12/2022 and 18/2022 — Guidelines on Section 194R
- SEBI BRSR Core — Business Responsibility and Sustainability Reporting
- NITI Aayog — Responsible AI for All (RAI) Principles