What AI can and can't do in Indian HR
Reviewed by Mellow Editorial Team, HR & payroll content team
AI can handle a defined set of repetitive HR tasks well — but Indian compliance is rule-dense, regionally variable and regularly amended, which creates hard limits on how far automation can go without human oversight.
What AI genuinely does well in HR
Repetitive, high-volume tasks are where AI earns its keep.
Screening and sorting. AI tools can scan hundreds of CVs against a job description, flag keyword matches and rank applicants. This saves real hours for a recruiting team. The caveat: the tool reflects whatever biases exist in the training data or the job description itself. Someone still needs to audit the shortlist.
Query handling. A well-configured HR chatbot can answer leave balance questions, explain policy documents, guide employees through a reimbursement process or tell someone which form to submit. These are high-frequency, low-complexity queries that consume a disproportionate amount of an HR team's time.
Drafting. AI writes a serviceable first draft of an offer letter, job description, HR policy or performance review template. The draft needs checking — especially anything that touches Indian law — but starting from a draft is faster than starting from blank.
Attendance and scheduling. Pattern recognition across attendance data, shift planning and flagging anomalies (unusual absenteeism, overtime trends) are all things rules-based and AI-adjacent tools handle reliably.
Where AI runs into Indian compliance limits
Indian statutory compliance is not a single uniform system. It layers central law, state-level rules, industry-specific schemes and frequent government notifications. This is where unchecked AI gets dangerous.
Payroll calculations involve multiple moving parts. Under the new income tax regime, slabs rise to 30%, with a section 87A rebate available for lower-income employees and a 4% health and education cess on top. Employees also contribute 12% of basic wages to EPF, with an equal 12% employer contribution. ESI applies separately for employees below the applicable wage threshold. A payroll engine can run these calculations — but someone has to confirm the inputs are correct, that the employee is on the right tax regime, and that any mid-year changes have been applied.
TDS, Form 16 and Form 24Q need human sign-off. TDS is deducted each month and deposited to the government. Employers file Form 24Q quarterly and issue Form 16 to employees after the financial year ends. An AI tool can flag mismatches and generate draft returns, but the authorised signatory is a person — and any error creates liability for the employer, not the software vendor.
Gratuity, notice pay and full-and-final settlements. Gratuity becomes payable after five years of continuous service. The calculation sounds straightforward, but disputes often arise around what counts as "continuous service", how to treat contractual staff, and whether last-drawn basic or gross salary applies. AI can apply a formula; it cannot exercise the judgment a well-briefed HR professional or labour lawyer brings to an edge case.
The four Labour Codes. India's four consolidated Labour Codes have been in force from 2025, but many state governments are still issuing their own rules under these Codes. The definition of "wages" under the Code on Wages, for instance, directly affects how EPF and gratuity are calculated. An AI tool trained on pre-2025 data or that does not track state-level notifications will give you an answer that is confidently wrong.
The specific risks of over-relying on AI in Indian HR
The liability stays with the employer, always. If a chatbot gives an employee wrong information about their tax liability, or if a payroll tool miscalculates EPF contributions, the company bears the legal and financial consequence. The vendor's terms of service almost certainly do not.
AI tools trained on global datasets often handle Indian-specific scenarios poorly — particularly for less common but perfectly legal arrangements like notice period buyouts, leave encashment taxation or the treatment of joining bonuses. These require someone who knows the rules.
There is also a data risk. Indian HR data includes Aadhaar numbers, PAN details, salary information and bank accounts. Feeding this into a third-party AI tool without understanding where the data goes and whether it is processed in compliance with India's data protection framework is a material risk, not a theoretical one.
How to use AI sensibly in Indian HR
Use AI as a drafter and a filter, not as a decision-maker or compliance authority.
The practical division: let AI handle the volume — the first-pass screening, the policy FAQ responses, the draft documents, the attendance dashboards. Keep a human in the loop for anything that creates a legal obligation, involves an employee's financial entitlements, or requires interpreting a rule that has changed in the last twelve months.
Audit AI outputs regularly. A tool that gave correct answers six months ago may not be current. Build a review cycle, especially around the start of each financial year when tax slabs, EPF wage ceilings and other thresholds are typically updated.
For companies running payroll across multiple countries or entity types, the principle is the same: automation handles scale, humans handle interpretation and accountability.
---
Run HR and payroll in India with Mellow
Mellow brings HR, payroll and 12 AI agents into one platform — built to handle India properly, with payroll included, from £4 per employee per month. The AI agents don't just answer questions; they generate contracts, run cost estimates and draft letters for you.
[Start a free trial →](/register)