TL;DR: AI makes the mechanical parts of contract work faster: first drafts, clause-library application, defined-term checks, and risk-allocation triage. The Indian-law fundamentals (free consent under the Indian Contract Act 1872, stamping under the Indian Stamp Act 1899, registration under the Registration Act 1908, and electronic signatures under the IT Act 2000) remain the lawyer’s responsibility to verify. AI can hallucinate clauses, cite wrong statutes, and miss state-level stamp duty. This post sets out a verified, end-to-end workflow so you get the speed without the risk.

On this page

Contracts are where legal risk first takes shape on paper. A well-drafted agreement allocates risk clearly, forecloses the disputes that eat client time and money, and does it in language that holds up if it is ever tested. Poorly drafted contracts do the opposite: they create ambiguity precisely where clarity was the whole point.

The problem is not that lawyers draft carelessly. It is that contract work is volume-intensive, deadline-driven, and structurally repetitive. A senior associate reviewing an incoming counterparty draft at 11 pm under a signing deadline is working in conditions that produce errors. A first-year being asked to produce a commercial services agreement from scratch is doing it from memory and precedents that may be outdated.

AI changes this calculus, not by replacing the lawyer’s judgment, but by handling the mechanical parts faster and more consistently than any human can. Used properly, AI compresses the grunt work so the lawyer can spend more time on the analysis that actually differentiates good advice from adequate advice.

This post covers the full workflow: from taking instructions to post-signature obligations, with the Indian-law fundamentals a drafter must keep in view at every stage.

The Indian Contract Act 1872: what every drafter must keep in view

Before a single clause is drafted, a lawyer needs to be clear about what makes a contract enforceable under Indian law. The foundational statute is the Indian Contract Act 1872, which sets out the requirements every agreement must satisfy to be treated as a binding contract.

Section 10: the basic test

Section 10 of the Indian Contract Act 1872 states: “All agreements are contracts if they are made by the free consent of parties competent to contract, for a lawful consideration and with a lawful object, and are not hereby expressly declared to be void.” This single sentence contains the four-part test that governs every commercial agreement you will draft.

Consent is not free if it has been caused by coercion (section 15), undue influence (section 16), fraud (section 17), misrepresentation (section 18), or mistake (sections 20 to 22). Section 14 defines free consent by exclusion: consent is free when it is not caused by any of those five vitiating factors.

The practical implications for drafting are real. A clause obtained under economic duress in a negotiation that approaches coercion may be voidable. A representation made during pre-contract negotiations that turns out to be false can expose the representor to liability under section 17 or 18. When you draft or review a contract, the background question is always whether consent was, and will remain, free.

Capacity: section 11

Section 11 specifies who is competent to contract: a person of the age of majority under the law to which they are subject, of sound mind, and not disqualified from contracting by any law. For corporate entities, capacity is governed by the company’s constitutional documents and the Companies Act 2013. A contract with a minor is void, not merely voidable, a distinction that matters when you are contracting with individuals whose age you have not verified.

Lawful consideration and lawful object: sections 2(d) and 23

Section 2(d) defines consideration: when, at the desire of the promisor, the promisee or any other person has done or abstained from doing, or does or abstains from doing, or promises to do or abstain from doing something, that act or abstinence or promise is called consideration for the promise.

Section 23 provides that the consideration or object of an agreement is unlawful if it is forbidden by law; is of such a nature that, if permitted, it would defeat the provisions of any law; is fraudulent; involves or implies injury to the person or property of another; or the court regards it as immoral or opposed to public policy. A contract with an unlawful object or consideration is void.

The practical implication: when drafting agreements in regulated sectors (financial services, pharmaceuticals, broadcasting, defence), check that the purpose the contract implements does not cross into territory the relevant regulatory statute treats as prohibited.

Breach and remedies: sections 73 and 74

Section 73 allows the injured party to recover compensation for any loss or damage caused by a breach that arose naturally in the usual course of things or was reasonably foreseeable by both parties when they contracted. Section 74 addresses liquidated damages: where a sum is named in the contract as the amount to be paid on breach, the party complaining is entitled to receive reasonable compensation that does not exceed that amount, whether or not actual damage is proved.

When you draft a liquidated damages clause, the cap matters. Courts have held that section 74 allows the court to award reasonable compensation up to, but not exceeding, the named sum. A punitive clause set far above likely loss may be moderated.

Electronic signatures and e-contracts under the IT Act 2000

Most commercial contracts today are executed electronically. The legal framework is the Information Technology Act 2000, as amended in 2008.

Section 5 of the IT Act 2000 gives legal recognition to electronic signatures. It provides that if any law requires a document to be signed, that requirement is satisfied by an electronic signature applied in the manner prescribed by the Central Government.

Section 10A, inserted by the 2008 amendment, provides that a contract formed through electronic means (where proposals, acceptances, and revocations are expressed in electronic form or by electronic records) is not unenforceable solely on the ground that electronic means were used.

This makes standard click-wrap and e-signed contracts legally valid for most commercial purposes. The critical exception is in the First Schedule to the IT Act, which excludes certain documents from the Act’s scope: negotiable instruments other than cheques, powers of attorney, trusts, wills, and contracts for the sale of immovable property. Those instruments require wet-ink signatures and, where applicable, registration.

For everything else (service agreements, supply contracts, NDAs, licence agreements, employment contracts) electronic execution is legally sufficient, provided all the section 10 requirements of the Indian Contract Act are met: free consent, competent parties, lawful consideration, lawful object.

Stamping and registration: the two procedural pillars

Stamp duty under the Indian Stamp Act 1899

Stamp duty is a fiscal imposition on certain instruments. Under section 35 of the Indian Stamp Act 1899, no instrument chargeable with duty shall be admitted in evidence for any purpose unless it is duly stamped. An instrument that is insufficiently stamped may be impounded. The deficiency can be cured by paying the outstanding duty plus a penalty (under section 35, the penalty is five rupees, or ten times the deficient duty, whichever is higher) but the cure takes time and creates practical complications in disputes.

The key drafting point: stamp duty rates are state-specific and vary by instrument type. The Indian Stamp Act 1899 sets central rates, but most states have enacted their own stamp legislation (or schedules under the central Act) that override or supplement the central rates for instruments executed within the state. An AI tool cannot reliably tell you the current stamp duty on a particular instrument in a particular state. That calculation requires checking current state-specific schedules, which change.

Compulsory registration under the Registration Act 1908

Section 17 of the Registration Act 1908 lists the categories of instruments that must be registered. These include instruments of gift of immovable property; non-testamentary instruments creating, declaring, assigning, limiting, or extinguishing rights in immovable property of value exceeding one hundred rupees; and leases of immovable property for a term exceeding one year.

An instrument that requires registration but is not registered is inadmissible in evidence of any transaction affecting immovable property and cannot be acted upon.

The drafting implication is straightforward: whenever a contract creates or modifies rights in immovable property, or is an instrument of gift of such property, registration is compulsory. AI can flag the issue; only a lawyer who has checked the current position for the relevant state and instrument type can answer definitively.

The end-to-end AI-assisted contract workflow

Stage 1: intake and instructions

Before directing AI to produce anything, establish the brief with precision. This means: the identity and capacity of the parties (including corporate authorisation); the subject matter and key commercial terms; the governing law and preferred jurisdiction; any specific statutory or regulatory requirements that apply; the client’s non-negotiable positions and fallback positions on key clauses; and the timeline for signing.

Vague inputs produce vague drafts. An instruction like “draft a service agreement” will produce a generic document. An instruction that specifies the services, the fee structure, the milestone-based payment trigger, the IP ownership position, the liability cap, the notice periods, and the governing law will produce something much closer to usable.

A good AI-assisted intake process often starts with a structured instruction template or term sheet that the lawyer fills in before the AI sees it. The AI’s output is only as precise as the instructions it receives.

Stage 2: first draft generation

For standard commercial arrangements (supply agreements, service contracts, non-disclosure agreements, licence agreements, distribution agreements) the structural logic is well-established. An AI tool can produce a working draft with the key clauses in the right order faster than even an experienced lawyer typing from a precedent library.

This is not the same as a finished draft. It is a starting point, one that is often structurally sound and captures the main commercial terms, but requires the lawyer to review it clause by clause before it goes anywhere near the other side. The value is that the blank-page problem disappears and the structural decisions are already made.

For complex, bespoke transactions (project finance agreements, joint venture constitutions, complex earn-out structures) the first draft from AI is likely to be a more distant starting point. It will get the structure right but will miss the deal-specific provisions that matter most. The lawyer’s role in refining those provisions is larger, not smaller.

Stage 3: clause library application

Firms and in-house teams with standard positions on commonly negotiated clauses (limitation of liability, indemnity, confidentiality, force majeure, intellectual property ownership) can use AI to apply those positions consistently across documents. Where the organisation has a preferred formulation, the AI can be directed to use it.

This is particularly useful for in-house counsel who draft high volumes of similar contracts and need language to be consistent across the portfolio, not just legally sound in isolation. A liability cap formulation that is appropriate for a small vendor agreement may be inappropriate for a strategic partnership, even if both agreements are governed by the same playbook. The AI applies the library; the lawyer decides whether the library position fits the deal.

Stage 4: risk review and redlining

When reviewing a counterparty’s draft, the AI workflow typically involves:

  • A completeness check: does the draft include all the provisions a careful lawyer would expect for this contract type?
  • A risk-allocation scan: are the key risk clauses (indemnity, liability cap, IP ownership, termination triggers) balanced, or do they systematically favour the counterparty?
  • A playbook comparison: where does the draft deviate from the organisation’s standard positions, and how significant are those deviations?
  • An internal-consistency check: are defined terms used consistently, do cross-references resolve correctly, and are there provisions that appear to contradict each other?

The AI produces a structured first read: a list of issues ranked by significance. The lawyer then works through that list, applies judgment to each issue (is this deviation commercially acceptable in this deal?), and drafts the redline.

Stage 5: negotiation tracking

Over the course of a negotiation, a contract accumulates tracked changes, counter-proposals, and agreed positions. AI can help manage this by comparing versions, identifying what changed between drafts, and maintaining a structured log of open issues versus agreed positions.

This is more organisational than analytical, but in complex transactions with many rounds of negotiation, having a reliable record of what is agreed and what is open saves time and prevents the embarrassing situation where a position thought to be agreed turns out to have been reverted in a subsequent clean draft.

Stage 6: pre-execution checks

Before the document executes, confirm:

  • Internal consistency: are defined terms used correctly throughout, do cross-references resolve?
  • Stamping: has the correct stamp duty been calculated for the instrument and the relevant state?
  • Registration: does this instrument require registration, and if so, has the process been initiated?
  • Execution requirements: for corporate parties, are the signing authorities correct, are board resolutions required, are company seals required?
  • Electronic signature: is electronic execution permitted for this instrument under the IT Act 2000 First Schedule exclusions?

Stage 7: post-signature obligations

Many contracts create obligations that survive execution and continue throughout the contract’s life: payment milestones, reporting obligations, renewal notices, termination windows, insurance requirements, compliance certifications. These are easy to miss once the signing pressure has passed.

AI can extract post-signature obligations from the signed contract and populate a structured obligations register: a checklist of who is required to do what, by when, and with what consequences for non-performance. For in-house teams managing a large contract portfolio, this is one of the most practically valuable uses of AI in contract work.

Where AI genuinely helps in drafting

Generating a first draft from a term sheet

The most immediate use is turning a term sheet, heads of terms, or a set of client instructions into a structured draft. For standard commercial arrangements, an AI tool can produce a working draft with the key clauses in the right order faster than typing from a precedent library. That speed is real. The blank-page problem goes away. The structural decisions (what goes in the recitals, where the definitions sit, how the operative provisions are sequenced) are already made.

Building and applying a clause library

Where the organisation has standard positions on commonly negotiated clauses, AI can apply those positions consistently across documents. Consistency matters in contract portfolios: a liability cap formulation that varies between identical transactions creates both legal uncertainty and negotiating disadvantage when counterparties compare notes.

Standardising language and catching defined-term drift

Inconsistent defined terms are a common source of contract disputes. A term defined one way in the recitals and used differently in the operative clauses creates genuine ambiguity. AI is reliable at identifying this kind of internal inconsistency: defined terms used before they are defined, definitions that do not match their usage, capitalisation that is applied unevenly. Catching this in drafting is vastly preferable to litigating it later.

Comparing a draft against a template or precedent

Many contracts start from a standard template that is then negotiated and marked up over several rounds. AI can accelerate the gap between “template” and “deal-specific first draft” by ingesting the term sheet and proposing where the template needs to change.

Where AI genuinely helps in contract review

Spotting missing or incomplete clauses

For a given contract type, there is a reasonably well-established set of provisions that a careful lawyer would expect to see. An AI tool can check whether those provisions are present and flag those that are missing or truncated. This is not a substitute for a lawyer reading the document. An AI can identify that a limitation of liability clause is absent; it cannot always assess whether its absence is commercially appropriate in a particular deal.

Identifying one-sided risk allocation

Reviewing a counterparty’s draft for risk balance (indemnities that are heavily one-sided, liability caps that only protect one party, IP assignment clauses that go further than the deal requires) is work that AI can triage. It will not always get the commercial context right, but flagging the provisions worth scrutinising saves the lawyer from starting that triage from scratch.

Comparing against a playbook

For in-house teams and firms that have formal negotiation playbooks (stating the organisation’s position on each negotiated clause, with fallback positions and absolute redlines) AI can compare an incoming contract against the playbook and surface every deviation. This turns what is often a slow, clause-by-clause cross-reference into something that takes a fraction of the time.

Catching drafting inconsistencies

The same internal-consistency checks useful in drafting are equally valuable in review. Defined terms used inconsistently, cross-references to the wrong clause number, provisions that appear to contradict each other: these are exactly the kind of structural errors that an AI scans for reliably and that a tired lawyer reading a long document can miss.

High-risk clauses to review with care

Not every clause in a commercial agreement deserves equal attention. These are the provisions where the stakes are highest and where AI-assisted review must be followed by the closest human scrutiny.

Indemnity clauses. An indemnity can transfer risk far beyond what a standard damages award would allow. Unlike a damages claim under section 73 of the Indian Contract Act, an indemnity may not require the indemnified party to prove loss. The scope of what is covered (third-party claims only, or also losses between the parties?), the carve-outs (negligence, wilful default), and the procedure for making claims all need careful reading.

Limitation of liability. These clauses cap what one party can recover from the other. Check whether the cap applies to both parties or only one; what categories of loss are excluded from the cap (typically: death, personal injury, fraud, wilful misconduct, and sometimes IP infringement); and whether the cap amount makes commercial sense relative to the value of the contract.

Termination provisions. Termination for convenience (allowing exit without cause on notice) and termination for cause (requiring a specified default) have very different risk profiles. Check the notice periods, the cure rights before a for-cause termination is effective, and the consequences of termination, particularly payment obligations for work done to the termination date, return of materials, and survival of obligations.

Governing law and jurisdiction. In a cross-border contract, the governing law determines which legal system’s rules apply to interpreting and enforcing the agreement. Jurisdiction determines which courts can hear disputes. These are not always the same, and the interaction between them can produce unexpected results. For Indian-law-governed contracts, specifying the seat of exclusive jurisdiction to the courts of a named city eliminates later arguments about which court has power to hear the claim.

Dispute resolution and arbitration. Where parties want to arbitrate, the clause must comply with the requirements of the Arbitration and Conciliation Act 1996 and clearly specify the seat of arbitration. Indian courts have examined arbitration clauses in detail, and poorly drafted clauses, particularly those that are ambiguous about seat versus venue, have generated significant satellite litigation. The seat determines supervisory court jurisdiction and the law governing the arbitration itself. When you rely on one of those decisions, it helps to know how to read and brief an Indian judgment so you extract the ratio rather than stray observations.

Confidentiality. Check the definition of confidential information (is it limited to information marked as confidential, or does it extend to anything a reasonable person would know to be confidential?); the exclusions (publicly available information, information independently developed); the permitted disclosures (legal advisers, regulators, group companies); and the post-termination survival period.

Force majeure. The list of triggering events matters. Pandemic, government action, cyber attack, supplier failure: each creates a different risk profile. Check whether the clause requires the affected party to mitigate and notify within a defined period, and whether prolonged force majeure entitles either party to terminate.

Intellectual property ownership and licensing. Where any work product, software, or creative output is being produced under the contract, confirm who owns the IP in that output. Absent a clear assignment clause, IP created by an independent contractor may remain with the contractor under the Copyright Act 1957. Where the contractor retains background IP, what licence is granted to the client, and on what terms?

Where AI is dangerous and where the human must decide

TaskAI can helpHuman must decide
Generating first draft from term sheetYes, structural first passWhether the draft captures deal-specific risk
Checking completeness against clause listYes, reliable completeness triageWhether a missing clause is commercially acceptable
Flagging risk-imbalanced clausesYes, surfaces deviationsWhether the imbalance matters given deal dynamics
Checking defined-term consistencyYes, very reliableWhether a definitional inconsistency is ambiguity or error
Calculating stamp dutyNo, state-specific, changesLawyer must check current state schedule
Determining registration requirementPartially, can flag potential needLawyer must verify against current statute
Arbitration seat vs venue draftingNo, nuanced, satellite litigation riskLawyer must choose seat and draft precisely
Sector-specific regulatory complianceNo, often incomplete or outdatedLawyer must verify against relevant regulation
Whether consent is free under section 14 ICANo, fact-specificLawyer must assess from instructions
Whether the object is lawful under section 23 ICAPartially, can flag obvious issuesLawyer must verify for the specific regulated sector
Post-signature obligations registerYes, extraction is reliableLawyer must confirm completeness and trigger dates
Negotiation strategy and positionNoLawyer (with client instructions)
Execution requirements for corporate partiesPartially, can flag generic requirementsLawyer must verify for specific corporate structure

Governing law, jurisdiction, and arbitration clauses

For contracts with any cross-border element (and for high-value domestic contracts where disputes are likely to be significant) the governing law and jurisdiction provisions deserve close attention.

Specify clearly whether Indian law governs. Where parties from different countries are involved, a governing-law choice eliminates the need to determine which country’s law applies under private international law principles, which is a litigation point in itself.

For domestic contracts between parties in different states, specify which courts have jurisdiction, and whether jurisdiction is exclusive or non-exclusive. A non-exclusive clause means that either party can also sue in any other court that has natural jurisdiction, which may not be the intended result.

Arbitration. Arbitration clauses are standard in most commercial agreements of any significance. Where parties want to arbitrate rather than litigate, the clause should comply with the requirements of section 7 of the Arbitration and Conciliation Act 1996. Section 7 requires the arbitration agreement to be in writing; it may be in the form of a clause in the main contract or a separate agreement.

Critically, specify the seat of arbitration, not merely the venue. The seat is a legal concept that determines the curial law and the supervisory jurisdiction of courts. The venue is the physical location where hearings take place. Indian courts have litigated this distinction extensively, and clauses that refer only to venue, or that designate a venue without making clear that it is also the seat, have produced jurisdictional disputes. The safe drafting practice is to specify the seat explicitly.

Also specify: the governing rules (institutional, SIAC, LCIA, ICC, MCIA, or ad hoc under the Act), the number of arbitrators, the language of proceedings, and, for international commercial arbitrations, whether Part II of the Act applies. These are commercial and legal decisions that belong to the lawyer and the client, not to the AI.

Sector-specific regulation and data protection

General-purpose AI output may not adequately reflect the regulatory requirements that apply to contracts in specific sectors.

Financial services. Contracts that involve the provision of financial services, lending, securities, or insurance engage the requirements of RBI or SEBI regulations. The terms of a loan agreement between entities subject to RBI guidelines, or a distribution agreement for mutual fund products, need to be checked against the relevant regulatory framework, not just general contract law.

Employment. Employment agreements engage multiple statutes: the Industrial Employment (Standing Orders) Act 1946, the applicable state Shops and Establishments Act, the Maternity Benefit Act 1961, the Payment of Gratuity Act 1972, and (where applicable) the Code on Wages 2019 and the other Labour Codes. AI may produce a stylistically correct employment agreement that is substantively non-compliant with the applicable labour statutes.

Data processing contracts. The Digital Personal Data Protection Act 2023 (DPDP Act) is now in force, though key substantive obligations are scheduled to take effect approximately 18 months from May 2027 based on notification timelines. Where a contract involves the processing of personal data, data fiduciaries are required under the DPDP Act to engage data processors “only under a valid contract.” This creates an obligation to include in service agreements, SaaS contracts, and outsourcing arrangements specific data processing provisions: scope of processing, security obligations, assistance with rights fulfilment, breach notification requirements, and data deletion on termination.

Check any technology contract, outsourcing agreement, or service contract involving personal data against the DPDP Act requirements. AI may produce a data processing clause that looks correct but does not reflect the specific obligations the Act imposes.

Healthcare and pharmaceuticals. Contracts for the manufacture, supply, or distribution of drugs and medical devices engage the Drugs and Cosmetics Act 1940 and the rules made under it, and (for medical devices) the Medical Devices Rules 2017. Compliance with those frameworks is not a contract-drafting question alone. It is a regulatory question that must be resolved before the contract is executed.

For related reading on how AI handles legal research accuracy, see our post on avoiding hallucinations in AI legal research.

The verification discipline: owning every clause

AI output in contract drafting and review is a starting draft and a structured first read. It is not a finished document and it is not legal advice. The distinction matters and should be kept clearly in view throughout the workflow.

The AI drafts; the lawyer decides. Every clause that goes out under your name is yours: the accuracy of the language, the appropriateness of the position, the legality of what it commits the client to.

This has practical implications.

Read every clause the AI produces. A clause that looks standard may not be appropriate for this particular deal, this particular client, or this particular counterparty. The AI does not know what the lawyer knows: the relationship history, the commercial context, the instructions the client gave in the meeting.

Cross-check the substantive law. AI models are trained on general legal knowledge. Their understanding of specific provisions of Indian statute, their application by Indian courts, and their interaction with sector-specific regulation may be incomplete or may not reflect recent developments. Where a clause engages a specific statutory provision, verify the position against the statute and relevant authorities yourself, ideally working from free authoritative judgments on e-SCR with neutral citations rather than the model’s recollection.

For guidance on how to verify that an authority you are relying on is good law, see our post on good law checking and citator tools in India.

Verify internal consistency with the rest of the negotiated agreement. A draft produced clause by clause, or assembled from multiple sources, may contain inconsistencies that were not visible in any individual pass.

Never rely on AI for stamp duty calculations or registration requirements. These are state-specific, vary by instrument type, and change. The consequences of getting them wrong (an inadmissible instrument under section 35 of the Indian Stamp Act 1899) are serious.

This is not a counsel of caution against using AI. It is the same discipline that a supervising partner would apply to a work product produced by a junior: read it, check it, and own it before it goes out.

Confidentiality when using AI tools with client contracts

When uploading contract documents to any AI tool, apply the same care you would apply to any external service handling client materials.

The Bar Council of India Rules of Professional Conduct impose obligations of confidentiality on advocates. Uploading a client’s draft agreement to a third-party AI platform without client consent, or without understanding how the platform handles the data, raises real professional conduct questions.

The minimum standard for any tool used in legal practice: client inputs are private, never sold, and never used to train public models.

Before deploying any AI tool in contract work, confirm:

  • Does the platform process your inputs on-premises, or does it send them to a third-party model API?
  • Is the data used for model training? If so, on what opt-out terms?
  • Where is the data stored, and for how long?
  • What happens if the platform is acquired or changes its terms?

For clients in regulated industries (banking, insurance, defence, healthcare) the data governance question may be governed by sector-specific rules as well as by professional conduct obligations.

A practical clause-review checklist

Use this checklist on incoming counterparty drafts. The “AI can flag” column indicates where AI is a reliable first-pass tool. The “human must verify” column indicates where the lawyer’s judgment is essential regardless of what the AI says.

ClauseWhat to checkAI can flagHuman must verify
PartiesCorrect legal names, capacity, authorisationPartialAlways, check company registration, signing authority
RecitalsAccurate, no unintended admissionsYesContext-specific admissions
DefinitionsConsistent with operative clauses, no circular definitionsYes, reliableWhether scope is commercially appropriate
ConsiderationStated, lawful, not illusoryYesWhether consideration satisfies section 23 ICA
IndemnityScope, exclusions, cap, procedureYes, flags asymmetryWhether scope is acceptable in this deal
Liability capAmount, exclusions, applies to both partiesYesWhether cap amount makes commercial sense
IP ownershipWho owns deliverables, background IP licenceYesCopyright Act 1957 implications
ConfidentialityDefinition scope, exclusions, post-term survivalYesWhether definition is appropriate for information exchanged
TerminationFor-cause triggers, cure rights, for-convenience noticeYesWhether notice periods are commercially workable
Force majeureTriggering events, mitigation obligation, termination rightYesWhether list fits deal-specific risk profile
Governing lawSpecified, consistent with jurisdictionYesChoice of law implications for enforcement
JurisdictionExclusive or non-exclusive, correct courtsYes, flags absenceWhether exclusivity is intended
ArbitrationSeat specified, rules specified, institutional vs ad hocYes, flags seat/venue ambiguitySeat choice, institutional selection
Stamp dutyWhether stamping is requiredFlags potential need onlyLawyer must check current state schedule
RegistrationWhether registration is requiredFlags potential need onlyLawyer must verify under Registration Act 1908
Electronic executionWhether excluded from IT Act 2000Flags First Schedule exclusionsLawyer must confirm for specific instrument
Post-signing obligationsPayment milestones, notices, renewalsYes, extraction reliableConfirm trigger dates and responsible party

Post-signature obligations and contract lifecycle

The tendency in contract practice is to treat signing as the end of the work. For commercial agreements of any duration or complexity, it is the beginning.

Post-signature obligations include: payment milestones and the invoicing procedures that trigger them; periodic compliance certifications; insurance maintenance and evidence requirements; notice periods before renewal or termination windows close; reporting obligations (financial, operational, regulatory); and audit rights.

Missing a termination notice window because no one tracked it is a common and costly error. The contract auto-renews at terms that may no longer be acceptable. Missing a compliance certification deadline may trigger a default. Missing an insurance renewal requirement may breach the contract’s obligations.

AI can extract these obligations reliably from the signed contract and populate a structured obligations register. For in-house teams managing a portfolio of contracts, an AI-generated obligations register (reviewed and confirmed by the lawyer before use) is one of the most practical tools available.

The broader point is that the contract lifecycle extends well beyond execution. AI supports the full lifecycle: drafting before signing, review during negotiation, and obligations tracking after signing.

Frequently asked questions

What makes a contract valid under Indian law?

Under section 10 of the Indian Contract Act 1872, a contract is valid when it is made by the free consent of parties competent to contract, for a lawful consideration and with a lawful object, and is not expressly declared void by the Act. Free consent, legal capacity, lawful consideration, and a lawful object are the four core requirements.

Section 14 of the Indian Contract Act 1872 defines free consent as consent not caused by coercion (section 15), undue influence (section 16), fraud (section 17), misrepresentation (section 18), or mistake (sections 20 to 22). If consent is vitiated by any of these factors, the contract is voidable at the option of the aggrieved party.

Are electronic contracts enforceable in India?

Yes, for most commercial contracts. Section 10A of the IT Act 2000 (inserted by the 2008 amendment) provides that contracts formed through electronic means are not unenforceable solely because electronic means were used. The First Schedule to the IT Act excludes certain documents: negotiable instruments (other than cheques), powers of attorney, trusts, wills, and contracts for sale of immovable property. Those require wet-ink signatures.

Does an NDA need to be stamped in India?

It depends on the state and the nature of the obligations. Many states treat a confidentiality agreement that creates an enforceable obligation as an instrument requiring stamp duty. If an NDA is unstamped or inadequately stamped, it may be inadmissible in evidence under section 35 of the Indian Stamp Act 1899. The safe practice is to check current state stamp schedules before executing.

What contracts must be registered in India?

Section 17 of the Registration Act 1908 requires compulsory registration for instruments of gift of immovable property; instruments creating, declaring, assigning, limiting, or extinguishing rights in immovable property of value exceeding one hundred rupees; and leases of immovable property exceeding one year. An unregistered instrument that requires registration is inadmissible in evidence of the transaction it records.

What happens if a contract has an unlawful object?

Under section 23 of the Indian Contract Act 1872, the contract is void. An agreement whose object is forbidden by law, defeats the provisions of any law, is fraudulent, involves injury to another person, or is immoral or opposed to public policy is unenforceable. Void means it is not merely unenforceable by the parties. It has no legal effect at all.

What should an arbitration clause include under Indian law?

A clause compliant with section 7 of the Arbitration and Conciliation Act 1996 must be in writing. Beyond that minimum, best practice is to specify: the seat of arbitration (not merely the venue); the governing rules (institutional or ad hoc); the number of arbitrators; the language of proceedings; and the governing law of the agreement. Ambiguity between seat and venue has generated significant litigation in Indian courts.

What is the difference between seat and venue in an Indian arbitration clause?

The seat is a legal concept: it determines the curial law (the law governing the arbitration process itself) and which courts have supervisory jurisdiction. The venue is the physical location where hearings take place. These can be different: a contract might designate Bengaluru as the seat (giving Karnataka courts supervisory jurisdiction) and allow hearings to take place anywhere. Confusion between the two has produced disputes over which court can entertain challenges to awards.

Can AI accurately calculate stamp duty for a contract?

No. Stamp duty rates are state-specific, vary by instrument type, and are amended periodically by state governments. An AI tool can flag that a given instrument is likely to be chargeable with stamp duty and may reference general central rates, but it cannot reliably calculate the correct duty for a specific transaction in a specific state. That calculation requires checking the current state stamp schedule.

What are the main risks of using AI for contract drafting in India?

The main risks are: hallucinated clauses that are plausible-sounding but legally incorrect; wrong statute citations (citing an English or American rule as if it were Indian law); missed state-specific stamp duty or registration requirements; failure to reflect recent statutory changes; and the application of general commercial law principles in situations that require sector-specific regulatory knowledge. The same failure modes appear in research, which is why the techniques for spotting hallucinated clauses and citations in AI legal research carry directly over to drafting. All of these can be managed with a proper verification discipline: reading every clause, cross-checking statutory references, and independently verifying stamp duty and registration requirements.

What Indian statutes does an AI-drafted contract most often get wrong?

Based on the structure of how general-purpose AI models work, the most common errors are in state-specific provisions (stamp duty, state amendments to central Acts), sector-specific regulatory requirements (labour codes, SEBI regulations, RBI guidelines), and recent legislative changes (the DPDP Act 2023, the 2019 and 2020 Labour Codes). These are exactly the areas where independent verification is most important.

How should I use AI to review an incoming counterparty draft?

Start by running a completeness check: does the draft contain all the provisions you would expect for this contract type? Then run a risk-allocation scan to identify imbalanced clauses. Then compare the draft against your playbook or standard positions. Finally, run an internal-consistency check for defined-term drift and cross-reference errors. The AI output is a structured first read; the lawyer then works through that list and applies judgment to each identified issue.

What is the AI-assisted contract workflow for an in-house team?

A practical in-house workflow: (1) complete a structured instruction template before engaging AI; (2) generate a first draft or run a first-pass review; (3) read the output clause by clause; (4) apply India-specific checks manually (stamp duty, registration, e-signature eligibility); (5) negotiate and redline; (6) run pre-execution checks; (7) extract post-signature obligations into a register. Steps 4 and 7 are where AI is most useful in the in-house context specifically.

Does the DPDP Act 2023 affect how I draft technology contracts?

Yes. Where a contract involves the processing of personal data, the DPDP Act 2023 requires data fiduciaries to engage data processors only under a valid contract. This creates an obligation to include specific data processing provisions in service agreements, SaaS contracts, and outsourcing arrangements. The full compliance timeline is tied to notification of rules, with substantive obligations set to take effect by approximately May 2027, but drafting these provisions now avoids retrofitting later.

What is a clause library and how does AI apply it?

A clause library is a curated set of preferred formulations for commonly negotiated provisions: the organisation’s standard indemnity language, preferred liability cap formulation, approved confidentiality carve-outs, and so on. AI applies the library by inserting the preferred language when generating a draft, or by flagging deviations from the library when reviewing a counterparty draft. The value is consistency across a large volume of contracts; the limitation is that the lawyer must still decide whether the standard library position is appropriate for each specific deal.

What post-signature obligations should be tracked in a contract management system?

At a minimum: payment milestones and the invoicing procedure that triggers payment; termination and renewal notice windows; insurance maintenance obligations and the deadline for providing evidence; periodic compliance certifications; reporting obligations (financial, operational, regulatory); and any regulatory approval conditions that must be satisfied post-signing. AI can extract these reliably from the signed contract; the lawyer should confirm the list before it becomes operational.

How does a force majeure clause work under Indian law?

The Indian Contract Act 1872 does not contain a statutory force majeure provision. Section 56 of the Act addresses subsequent impossibility (frustration), but commercial force majeure clauses are contractual, not statutory. The clause suspends performance obligations during a defined triggering event, typically requires prompt notification and mitigation, and may allow termination if the event exceeds a specified duration. The definition of triggering events is negotiated, and what counts as force majeure depends entirely on the clause’s language, not on any statutory list.

What is the difference between an indemnity and a damages claim under section 73?

A damages claim under section 73 of the Indian Contract Act 1872 requires the claimant to prove loss that arose naturally from the breach or was reasonably foreseeable. An indemnity is a contractual obligation to hold the indemnified party harmless from specified losses, which may arise from third-party claims and which may not require the same proof of causation or foreseeability. Indemnities can therefore be significantly more powerful than a statutory damages claim, which is why their scope and exclusions need careful drafting and review.

Should I use institutional or ad hoc arbitration rules in India?

This is a legal and commercial decision. Institutional arbitration (under SIAC, ICC, LCIA, MCIA, or DIAC rules) provides procedural infrastructure, an appointing authority if parties cannot agree on arbitrators, and administrative support. Ad hoc arbitration under the Arbitration and Conciliation Act 1996 gives parties more flexibility but requires more self-management. For high-value or cross-border disputes, institutional rules are generally preferred because they reduce the risk of procedural disputes about the arbitration itself consuming as much time and cost as the underlying dispute.

What should I tell clients about AI-assisted contract work?

Be transparent: AI tools assist in producing and reviewing drafts, but every clause in the final document has been reviewed and approved by a lawyer. The output is the lawyer’s work product, not the AI’s. Clients should also understand that their contract documents are handled under the same confidentiality obligations as all client materials, and should know what data governance standards the tools used in their matter meet.

Drafting and reviewing with Indian-law grounding

Contract work is exacting because the stakes are exacting. An error in a draft costs very little to fix at the drafting stage and very much to fix after execution, and more still to litigate. AI does not change that calculus. It changes how fast a lawyer gets from a blank page to a rigorous first draft, and how quickly the first-pass review of an incoming document surfaces the provisions that need attention.

Used with the verification discipline the work demands, that speed is real and meaningful: more time for the analysis and negotiation that require a lawyer’s judgment, less time on the mechanical production that does not.

Niyam is built for Indian legal practice, with a corpus grounded in 72,000+ Indian judgments and statutes, and citations you can verify. Its AI contract drafting and review tools generate first drafts, apply clause libraries, and run completeness and risk-allocation checks against an Indian-law-grounded corpus. Client inputs are private, never sold or used to train public models. You can start with a refundable ₹100 hold and try it at app.niyam.ai.