How to Patent an App Idea in India (And Why You Mostly Can't)

"An idea" is not a thing you can patent — anywhere

Every few weeks a founder asks us some version of the same question: "I have an app idea. How do I patent it before I tell anyone?"

The honest answer is that you can't, and the reason isn't bureaucratic sludge. A patent is granted for an invention — a specific technical solution, disclosed in enough detail that a skilled engineer could rebuild it from your filing alone. If you can describe the whole thing in a paragraph over coffee, you don't have a patent application. You have a product spec.

That distinction is not a formality you can drafting-trick your way past. It's the deal patents are built on: the state gives you a 20-year monopoly, and in exchange you publish the invention so everyone else can learn from it. No disclosure, no patent.

And in India there's a second wall behind that one, which is where most app ideas actually die.

What Section 3(k) actually blocks

Section 3(k) of the Patents Act, 1970 excludes from patentability "a mathematical or business method or a computer programme per se or algorithms" (WIPO's record of the provision).

Read that list slowly, because it covers almost everything a typical app is made of. Business method: excluded. Algorithm: excluded. The software itself: excluded, per se.

The entire fight over software patents in India happens inside those two Latin words.

The "per se" gap and the technical effect test

"Per se" implies the exclusion isn't total — a computer programme by itself is barred, but an invention that happens to use software isn't automatically dead. The question is whether you're claiming the program, or claiming a technical solution to a technical problem that the program happens to implement.

The anchor case is Ferid Allani v. Union of India (Delhi High Court, 2019), where the court held that computer-related inventions showing a technical effect or technical contribution can be patentable even though software is involved. The judgment offered concrete examples of what counts: higher processing speed, reduced hard-disk access time, more economical use of memory, a more efficient database search strategy (SFLC's analysis of the judgments that shaped the technical effect doctrine).

Notice what all of those have in common: they are improvements to how the machine works, not to how the business works. That's the line.

The business method trap — where most app ideas die

Here's the part founders consistently miss, and it's the single most important paragraph in this post.

The technical effect argument cannot rescue a business method. The Delhi High Court has held that the bar on business method patents is an absolute bar — the Patent Office doesn't even get to the technical effect analysis. The only question is whether the application addresses a business or administrative problem and offers a solution to it. If yes, it's refused, full stop (AIPPI's analysis of the Comviva Technologies decision; see also the European Commission's IP Helpdesk summary).

So: a marketplace that matches buyers to sellers in a novel way. A new subscription and pricing model. A smarter loan approval flow. A workflow that removes three steps from an insurance claim. These are business solutions to business problems. They are not patentable in India, no matter how clever, how novel, or how much revenue they generate — and no attorney can draft around it.

That is the ceiling for the overwhelming majority of app ideas. It's worth knowing on day one rather than after ₹1.5 lakh of legal fees.

To be fair, the Delhi High Court itself has flagged that this blanket exclusion deserves a re-look, noting that many inventions from startups, SMEs and educational institutions now fall in exactly this space. But that's a signal about where the law might go, not a strategy you can file on today.

Three things that changed in 2025–2026

This area moved more in the last eighteen months than in the previous eight years.

1. New CRI Guidelines (July 2025)

The Patent Office issued fresh Guidelines for Examination of Computer Related Inventions, 2025, replacing the 2017 version. They were released on 29 July 2025 after two rounds of public drafts (PIB press release; full guidelines PDF%20-%202025.pdf)).

What matters for you:

• A step-wise assessment methodology for Section 3(k), with flowcharts — examiners now have a visible rubric, which makes outcomes more predictable.
• A dedicated chapter on AI, machine learning, deep learning, blockchain and quantum computing, with scenario-based examples and explicit sufficiency-of-disclosure expectations.
• 20 worked examples in the body plus 40 more in an annexure, showing allowable versus non-allowable claims.
• Novel hardware is not mandatory to escape the exclusion — an old myth, now formally dead.
• Business methods remain absolutely barred.

If you're seriously considering a filing, have your attorney walk you through the annexure examples closest to your invention. It's the cheapest hour of diligence available.

2. Your screens just became protectable (March 2026)

On 9 March 2026, the Calcutta High Court ruled in a batch of appeals led by NEC Corporation v. Controller of Patents and Designs that a graphical user interface can qualify as a registrable "design" under the Designs Act, 2000. The court held there is no per se bar, that "article" and "design" must be read broadly and purposively, and that visibility during normal use is what counts — setting aside the Controller's rejections and remanding them (SCC Online report; case details via Verdictum).

This is not a patent. It's a design registration, it protects appearance rather than function, and it's far cheaper and faster. For a consumer app whose actual moat is a distinctive interface, this is more relevant than anything in the Patents Act. Caveat worth knowing: animations and flows can't be registered as moving designs — you file a series of static states.

3. The government stopped paying your patent attorney (March 2026)

For a decade, the SIPP scheme did something unusually generous: the government bore the entire fee of empanelled facilitators — patent attorneys — for any number of filings by a recognised startup. You paid only statutory fees (Startup India's IPR page; scheme document).

The notified version of that scheme expired on 31 March 2026, with no announced successor. Startups that hadn't begun filing before that date can no longer access it, and professional costs now land on you — roughly ₹30,000 to ₹80,000 or more per application depending on complexity (Chambers and Partners analysis).

Do not confuse the two things that are often mixed up here. The 80% statutory fee rebate for startups and small entities lives in the Patents Rules, not in SIPP — it survives. What ended was the free attorney. Since drafting is where nearly all the real money and nearly all the quality goes, the economics of a startup patent got meaningfully worse this year.

A blunt test: does your app contain an invention?

If every row that describes your product says "No", that isn't a failure. It means your defensibility lives somewhere other than the Patent Office, and you should stop spending on this and go build.

The protection stack that actually applies to apps

On copyright: software is a "literary work" under the Copyright Act, 1957, and protection arises automatically the moment the code is written — no filing needed. India follows the Berne Convention, so that protection is recognised across essentially every major software market without local registration. Registration is still worth it because the certificate is prima facie evidence of ownership in court, which matters enormously in a dispute, a due diligence, or an acquisition. Software registration requires filing source and object code with the application (Copyright Office FAQ; official fee details).

On trademarks: your app name is the asset users actually type. Class 9 covers downloadable software, Class 42 covers SaaS and software services, and most apps need both — one Form TM-A can cover multiple classes, with a fee per class (IP India's official trade mark fees). Recognised startups get a 50% rebate. This is the highest-return IP rupee a founder spends, and it's the one most often skipped.

If you genuinely do have an invention: the sequence

1. File a provisional specification first. It's cheap, it locks your priority date, and it buys you 12 months to build, test the market, and decide whether the full spend is justified.
2. Watch the 12-month cliff. Under Section 9(1), if the complete specification isn't filed within 12 months, the application is treated as abandoned and the priority date is gone permanently — there's no revival. This is the most common own-goal in Indian patent practice.
3. File the complete specification with claims. This is where drafting quality decides everything.
4. File the Request for Examination. The Patents (Amendment) Rules, 2024, effective 15 March 2024, cut this deadline from 48 months to 31 months from the earliest priority date. If you learned this process before 2024, your mental model is wrong — you no longer have a buffer.
5. Consider expedited examination under Rule 24C, which recognised startups qualify for. It pulls the first examination report forward to roughly 1–3 months instead of 12–24.
6. Going global? The PCT route gives you 31 months from priority to enter national phase in India (WIPO PCT Applicant's Guide, India). Miss it and rights are lost irreversibly.

Official fees are published in the First Schedule — the ₹1,600 e-filing fee covers up to 30 pages and 10 claims, and a startup pays it where a large company pays ₹8,000. Extra pages and claims cost more. To claim the concession you need DPIIT recognition and the supporting form.

The risk nobody warns non-technical founders about: you may not own your code

This is the one that actually destroys deals, and it has nothing to do with patents.

Under Section 17 of the Copyright Act, the author is the first owner — the person who wrote the code. Employment under a contract of service is an exception: your employer owns what you write on the job. But an independent contractor, freelancer, or outsourced agency is not an employee. Absent a written assignment, the developer owns the copyright in your product, not you (Section 17, India Code).

And Section 19 requires assignment to be in writing and signed by the assignor, identifying the work, rights, duration and territory. Get sloppy and the statutory defaults bite: no duration stated means the assignment lapses in five years; no territory stated means it covers India only.

Sit with that. A founder who paid an agency ₹40 lakh, has no signed assignment, and is now in a funding round can discover that their core asset reverts in five years and was never theirs outside India. We have seen versions of this in diligence. It is entirely preventable and entirely unfixable after the developer stops returning calls.

So: before you spend a rupee on patent counsel, make sure every contract with every person who has touched your codebase contains a written, perpetual, worldwide assignment. That's the cheap insurance. The patent is the expensive lottery ticket.

This is also why we're blunt about IP in the way we structure work. Under our Build-Operate-Transfer model, clean ownership isn't a clause you negotiate at the end — the whole point is that the product, the code, and the rights land with you at transfer. If you're weighing an agency, a freelance team, or a BOT partner, ask each one exactly who owns the copyright on day one and what happens if the relationship ends tomorrow. The answers differ more than the engagement models suggest.

What we tell founders

India is filing more patents than ever — 110,375 applications in 2024–25, up nearly 20%, with domestic filings now 61.79% of the total (IP India Annual Report 2024–25). Indian applicants grew filings 19.1% globally in 2024, a sixth straight year of double-digit growth, and the resident share has climbed from 28.1% in 2014 to 60.1% in 2024 (WIPO World Intellectual Property Indicators 2025). IP filings overall are up 44% in five years (PIB).

That boom is real, and almost none of it is app founders patenting app ideas. It's pharma, chemicals, electronics, manufacturing, and university research.

Our honest position, after building products for founders who arrive convinced the patent is step one:

The patent follows the build, never the other way round. You cannot draft a sufficient disclosure for a system you haven't built. Founders who try to patent first spend six months and real money describing a machine that doesn't exist, then build something different anyway — and the filing protects the thing they abandoned.

Your moat is almost certainly not legal. For consumer and SME software it's distribution, speed of iteration, the data you accumulate, switching costs, and knowing your domain better than anyone. A patent that takes three years to grant and lapses without renewal protects a version of your product that will be two rewrites obsolete by the time it's granted.

Secrecy is over-rated as a strategy and under-rated as a discipline. Nobody is going to steal your idea — execution is the hard part, and the people capable of executing it have their own ideas. But do keep your unpublished technical specifics unpublished, and do use NDAs with vendors. Note the interaction: publishing or demoing your invention before filing can destroy novelty. If you have a genuine invention, file the provisional before the launch, the demo day, or the conference talk.

If you're deep tech, file. If your product's value genuinely lives in a novel technical method — a signal processing advance, a materials-adjacent process, a real architectural innovation with measurable machine-level effects — then patents matter, investors will ask, and the SIPP expiry means budget ₹50,000–₹1,00,000 per application in professional fees that used to be free.

Do this month

• Trademark search and file your app name in Classes 9 and 42. Highest return, lowest cost, most-skipped.
• Audit your contracts for written copyright assignment from every developer, designer, and contractor. Fix the gaps before your next raise.
• Register copyright in your codebase once it's stable. Cheap evidence.
• Screenshot-file a design if your interface is genuinely distinctive — the March 2026 ruling just made this worth doing.
• Ask one honest question about the patent: does my product improve how the machine works, or how the business works? If it's the business, close the tab and go get users.

If you're trying to work out whether what you're building has a real invention in it — or you just want the thing built and owned cleanly — tell us what you're working on. We'd rather talk you out of a patent you don't need than sell you one you'll never enforce.

---

This post is general information, not legal advice. For a specific filing decision, work with a registered Indian patent agent or IP attorney.