Wages & benefits
Code on Wages — minimum wage, equal pay, timely payment
India's four Labour Codes came into force on 21 November 2025. Karya Vaani turns that regulatory shock into an operating system: it demystifies the obligation, structures the decision before it becomes a liability, and reaches every worker — direct or contract — in a language they actually understand.
The Readiness Survey is the honest opening question — not a sales pitch. Across manufacturing, construction and logistics, the same five pain points surface. The four codes did not simplify obligation; they consolidated it, raised the penalties, and made the principal employer answerable for the contractor's lapses.
"Are all my contract workers — including those under sub-contractors — correctly in EPFO and ESIC?" Most leaders cannot answer with confidence. The liability is theirs anyway.
Wage registers, induction records, licence tracking, incident logs — performed manually, with no single system of evidence and no audit-grade trail.
A plant manager needs 40 workers fast. The compliance impact of "permanent vs. contract vs. fixed-term" is evaluated after the threshold is already crossed.
Safety briefings, consent notices and alerts are issued in English. The workers who must act on them think in Telugu, Hindi and a dozen other languages.
The codes only feel impenetrable because obligation, threshold and liability live in separate documents. Karya Vaani collapses them into one structured rule library — each rule carrying its source, jurisdiction and last-verified date.
The s.2(y) wage definition forces basic pay to ≥50% of CTC — quietly the largest financial-impact item, lifting PF, gratuity and bonus liabilities the moment it's operative.
Standing orders, fixed-term employment and pro-rata gratuity (s.53(2)) — reshaping how provisioning and accruals are calculated for every FTE.
EPF, ESI and gratuity — where the contractor blind spot turns into principal-employer joint liability the instant a deployed worker is found uncovered.
Appointment-letter mandate, women on night shift consent (s.43), and the safety-briefing rules that gate floor access. High audit-failure surface.
Engage labour through a third-party contractor and the statutory obligation does not stop at their gate. If Sri Lakshmi Engg under-enrols 4 of 142 deployed workers in ESIC, the ESIC challan shortfall — and the penalty — attach to you, the principal employer. Karya Vaani exists to make that exposure visible before it crystallises.
Karya Vaani is not a checklist. It's an architecture: pillars that do the operational work, a localisation engine that removes the friction of reaching workers, and a governance layer that makes every action provable.
Where the compliance work actually happens — from structuring a hire to reconciling contractor ESIC to detecting a safety pattern.
The shared layer that takes anything the pillars produce and delivers it — text and voice — in the worker's own language, over WhatsApp, with proof of receipt.
The trust spine — inspectable rules, gated multilingual publishing, an append-only audit trail, and a clean handoff to your systems of record.
This is where the plant head and CHRO start every morning. It synthesises live signals from talent acquisition, contractor compliance and OHS into a single statutory posture — and, critically, every headline number opens to the exact formula and weighted components that produced it. No black box; a number you can defend to the board or an inspector.
Tap any ⓘ and the metric unpacks. The compliance score, for example:
Each component is an expected value — potential penalty × probability of enforcement. Statutory penalty ₹0.4–0.9 Cr (the s.2(y) wage-restructuring shortfall is the single largest driver), operational-stop risk ₹0.2–0.7 Cr, customer-audit failure ₹0.2–0.7 Cr.
Throughput = direct (~38/wk, 27%) + contract (~104/wk, 73%) completions on a 4-week rolling average — "contract intensity rising" warns when contract outpaces direct. Gaps = 4 critical (hard-gate push-to-HRIS, carry joint liability) + 19 advisory, rolled live from Vendor compliance.
Live status across every module. The composite score is never abstract — click a row and you land in the module that owns the problem. This is how plant management goes from "we're at 72" to "the Compressor Line ESIC mismatch is the thing to fix today."
Annualised cost-to-company of this cycle's hires against the sanctioned ₹80 L budget (₹46 L direct + ₹34 L contract). Variance isn't a lonely number — click a group and the excess breaks down to the specific Position IDs and contractors that drove it, each tagged to the decision that approved it.
The insight the budget bar gives plant management: the contract line is over not by accident — the excess maps to two work orders engaged against decision KN-2026-037 at higher-than-planned skill mix. Because every hire is tied back to an approved Karya Nirṇay decision, an overspend is always answerable: which requisition, which contractor, which approval.
Each pillar carries the same three-part promise: a value proposition, an AI-driven workflow doing the heavy lifting, and analytics that drill to the contractor and the process.
Structure the hire before it becomes a liability. A guided 4-step builder maps a workforce requirement against all four codes and recommends the structure that is both legally compliant and commercially optimal.
The rubric is hand-tuned and inspectable; the engine evaluates 5 workforce structures against the live rule bundle and surfaces an explained recommendation — never a black-box verdict.
Per-decision threshold map; committed annual cost vs. budget variance by hiring group, with an excess-over-budget drill-down.
An approved decision (KN-2026-037) becomes the only requisition onboarding can act against — tagging contractor, CLRA and category automatically downstream.
Track every Position ID through an inspectable progression ladder — success factors, SLAs, breach reasons and ownership captured at every stage.
SLA-breach prediction and stage-ageing detection nudge recruiters before a requisition stalls; ownership is always a named human.
Time-in-stage and breach reasons rolled up by function, location and recruiter — bottlenecks become visible, not anecdotal.
Each requisition is a persistent Position ID; the ladder is the evidence that the hire was governed, not improvised.
Verify, induct and attest — differently for direct vs. contract. One worker entity, a worker_type discriminator, shared document store and audit trail; statutory paths diverge only where the law demands.
Document extraction at intake — UIDAI eKYC with OTP consent, auto-crop to ID standard, ML field extraction — collapses manual data entry and its errors.
Onboarding throughput (142 / 30d · 94 contract · 48 direct) and document validity tracked as evidence-grade, not status text.
Universal appointment-letter rule met for every new joiner; contract workers carry CLRA-rights and ESIC-card attestations the law requires of them specifically.
Get every joinee floor-ready with PPE sizing, fitment and statutory briefings — where a safety briefing cannot be marked complete without PPE issued.
Modules localised by VAANI into TE / HI / EN automatically; competency tracks routed by role and zone.
Certificate validity at 100% evidence-grade; module sign-off tracked at section and quiz-question level.
Hard interlock to Factories Act s.41-B & Schedule 21 — the PPE-gate makes "trained on paper, exposed on the floor" structurally impossible.
A real-time view on contractor compliance. Contractors self-register; you approve and get a live score, ESIC/PF reconciliation against actual deployment, and quantified joint-liability exposure.
Contractor data anomaly detection — reconciles ESIC/PF challans against deployed headcount and flags the shortfall the moment it appears.
Live per-contractor compliance score, CLRA licence-expiry countdown (3 renewals due in 30d), and exposure quantified in rupees.
Sri Lakshmi Engg → ESIC covers 138/142 · CLRA expires in 21d · score 42 (red)
That single line is the difference between knowing and discovering at inspection.
Patterns are louder than incidents. A single safety ledger cross-references incidents and near-misses with worker presence and time — so same-zone, same-shift, same-contractor clusters surface as patterns, not isolated entries.
Pattern detection clusters the ledger across zone, shift, time-of-day and contractor; a detected pattern auto-routes a zone refresher and cues a broadcast.
Near-miss heat by zone; contractor incident-share vs. headcount-share; Form 21 / ESIC Form 16 statutory filing clock.
One detected pattern dispatches a localised refresher to 38 workers (24 direct · 14 contract) in TE/HI/EN — closing the loop the same shift.
Posture, exposure, and a sequenced roadmap. Rolls up signals from every pillar plus standalone statutory tracks — every recommendation tagged with the applicable regime by state and a last-verified date.
NLP gazette monitoring keeps the rule bundle current; inspector-letter triage routes incoming notices to the right owner.
Coverage matrix across AP / TN / Gujarat; exposure band ₹40 L – ₹1.6 Cr; high-impact gaps ranked by financial weight.
Contractor & CLRA domain scored weakest (54) — and the dashboard says why: it's driven by the open Vendor-compliance gaps. Exportable 90-day plan.
This is the friction-killer, and it is the part of Karya Vaani that most platforms simply don't have. A migrant workforce doesn't read English safety notices. The Localisation Engine takes anything the pillars produce, renders it in the worker's own language as both text and a voice note, delivers it over WhatsApp, and tracks it to acknowledgement.
The author writes once in English; VAANI localises. The worker roster on a single AP site already spans Telugu, Hindi, Tamil, Odia and Bengali — the exact migrant mix the ISMW Act governs.
Pick exactly two Indian languages per broadcast; every recipient is delivered text and a voice note in their own. Karya Vaani AI handles both the translation and the voice.
A worker who can't read a printed notice can listen to a 20-second voice note in their mother tongue while gloved-up at a machine. Voice is what makes "I issued the alert" become "the worker received and understood it."
Each worker sets a personal language preference from their own home screen; Karya Vaani delivers every future message in that choice. Personalisation isn't assigned top-down — it's owned by the person who has to act on the message.
No app to install, no portal to learn. The Karya Vaani assistant appears as a normal WhatsApp contact. Here is a live weather-stoppage safety alert as it lands on a Telugu-speaking worker's phone — text, a tappable voice note, and a one-tap acknowledgement that becomes audit evidence.
Every other compliance tool stops at "broadcast sent." Karya Vaani goes three steps further, and each step is a piece of defensible evidence:
Compose once, localise, broadcast — and prove it.
Every alert, in their language. The assistant appears to each worker as a normal WhatsApp contact.
A published weekly plan — 5 buses, 5 zone routes, two shifts, each route with its pickup points and timings — pushed to workers as a localised broadcast in one click. But the real value is what happens between publications.
The change reaches the 30-odd workers on that route, not the whole plant — so the signal isn't lost in noise and people actually act on it.
Acks tell the supervisor who has seen the change before the bus rolls — a missing worker is visible while there's still time to call.
A pickup confusion that used to cost a half-shift of no-shows becomes a 90-second broadcast — directly protecting line staffing.
This is not generic compliance content. It is this plant's body of knowledge — its processes, standards, policies, SOPs and training material — authored once by the customer and then localised and delivered through the same engine that runs every broadcast, so the floor learns and gets alerted in one consistent voice.
Consistency: the words used to train a worker on the chemical SOP are the same words used to alert them when a pattern triggers a refresher — no drift between the manual and the message.
Language of choice: authored in English, delivered in the worker's own language, with the localisation governed and audited — improved communication that's also defensible.
Closed loop: when OHS detects the Paint Shop Zone-3 pattern, it auto-routes the relevant module from this library to the 38 affected workers — not a generic notice.
Most compliance tools are single-sided: an HR admin console, and everyone else is a passive recipient. Karya Vaani is three-sided. The CHRO has her dashboard — but the worker and the contractor each get their own home screen where they can act, not just be tracked. This is where compliance stops being something done to people and becomes something they participate in.
Acknowledging an alert, reporting a hazard, or switching language is one tap. The worker becomes an active node in the compliance loop, not a name on a non-response list.
The labour supplier sees their own score, their own exposure, and the exact actions to lift it — turning a quarterly audit fight into a continuous, self-served cleanup.
Priya stops chasing and starts steering: every surface feeds her dashboard, so she manages by exception against the few signals that are actually red.
Most compliance dashboards stop at a number on a wall. Karya Vaani treats every broadcast and every signal as a measurable event with a target, attributes the misses to a named cohort, and hands the user a specific, dated action. The flagship example is the broadcast analytics engine, built around four objectives.
A safety alert isn't "sent" — it's received, over a window. The engine measures that window so it can be shrunk.
The same window, re-cut every way that lets you act on it — so "12% haven't acknowledged" becomes "the night shift on the Compressor Line, mostly Odia-speaking contract workers."
Time-to-read split by message criticality — a fire-safety alert and a transport-schedule update are held to different windows, and tracked against different expectations.
The four cuts auto-generate recommendation cards — each a specific, dated change the broadcaster can make to shrink the window or lift the response rate. Not "engagement is low," but "re-send the Odia voice note to the 14 non-acknowledged contract workers on the 14:00 shift by Thursday." Analytics that end in an action, with an owner and a date.
Incidents scored across zone × shift × time-of-day × contractor. Output isn't "17 near-misses" — it's "Paint Shop Zone-3, post-meal window, Sri Lakshmi workers at 35% on 11% headcount," each with a routed action.
Posture rolled into a banded estimate — ₹40 L–₹1.6 Cr statutory, ₹2.4 Cr contract value at risk, ~12% audit-failure probability — with high-impact gaps ranked by financial weight, not alphabetically.
ESIC/PF reconciled against deployment into a live per-contractor score, CLRA expiry countdowns, and joint-liability exposure — every input inspectable beneath the number.
Compliance software that can't prove its own decisions is just another spreadsheet. Governance is where Karya Vaani earns the right to be the evidence of record.
Statutory rules + layered customer policy. Each carries source, status, jurisdiction, last-verified date. Bundles are signed before activation — a transparent rubric, not an AI black box.
High-stakes localised artefacts route to an appointed reviewer for inline edit and approval. Reviewer identity + timestamp are written to the trail; content is then locked for distribution — accuracy is governed, not assumed.
Every translation, broadcast, ack, upload and rule activation is an immutable, SHA-256-chained, WORM entry — retained per statutory floor on your platform.
On onboarding, the worker master pushes to your HRIS / EHS / payroll over mTLS + HMAC-SHA256 signed, idempotent webhooks. Karya Vaani's responsibility ends cleanly at your gate.
Karya Vaani AI is applied where it removes real friction and toil, and deliberately withheld where a decision must be inspectable. The scoring rubric itself is hand-tuned and auditable — the AI assists the judgement, it does not hide it. There are five bounded jobs it does.
Continuously reads statutory gazettes so the rule bundle stays current — the difference between "compliant as of last quarter" and compliant today.
eKYC + field extraction kills manual data entry and the transcription errors that quietly seed compliance gaps.
Classifies and routes incoming statutory notices to the right owner, so nothing decays in an inbox until the deadline passes.
Reconciles ESIC/PF against deployment and flags the shortfall the instant it appears — the engine behind the Vendor-compliance score.
Indian-language translation and voice — the friction-killer that puts every alert, briefing and consent notice into a language the worker actually thinks in.
The discipline: AI does the reading, extraction, triage and translation — the toil. The scoring rubric, the rule sources and the recommendations stay hand-tuned, cited and inspectable. When the platform says your contractor domain scores 54, you can open the rules and the signals that produced it.
The platform's value is only real when it lands on a name and a workflow. Here is the same engine seen from the contractor level and the process level.
Signal 1 · Vendor compliance: ESIC covers 138 of 142 deployed workers → critical gap, principal-employer exposure attached.
Signal 2 · same source, OHS: their workers account for 35% of incidents on 11% of headcount → cluster flagged.
Signal 3 · Licence clock: CLRA licence expires in 21 days → renewal countdown live.
Result: one contractor, three independent signals, one compliance score of 42 (red) — surfaced weeks before any of it would have shown up at an audit.
Detect: OHS pattern detection clusters 4 near-misses in Paint Shop Zone-3 within 11 days.
Route: a zone-specific refresher is auto-dispatched to the Knowledge Center for 38 affected workers.
Reach: VAANI localises it to TE/HI/EN — text + voice — and broadcasts over WhatsApp.
Prove: reads and acknowledgements land in the SHA-256 audit trail as evidence.
Result: incident → corrective action → proof of delivery, with no spreadsheet and no language barrier in the loop.
Take the Readiness Survey for a scored sector benchmark; request a Karya Niyam gap report mapping every applicable 2026 rule to your answers; run a live hire through Karya Nirṇay to see compliant-and-cost-optimal in one screen.