Next due
20 OCTGSTR-3B · Summary return · Sep 2026in 4 days 21 OCTTax Audit Report · Form 3CA/3CB · AY 2026-27 · extended from 30 Sepin 5 days 30 OCTAOC-4 · Financial statements · FY 2025-26in 14 days 7 NOVTDS / TCS deposit · Deducted in Oct 2026in 22 days 11 NOVGSTR-1 · Outward supplies · Oct 2026in 26 days 15 NOVPF & ESI · Contributions · Oct 2026in 30 days 21 NOVITR filing · Audit cases · AY 2026-27 · extended from 31 Octin 36 days 29 NOVMGT-7 / 7A · Annual return · FY 2025-26in 44 days
All due dates
Patent Live

CRI Guidelines 2025, Annexure I: Examples 35 to 40 on the algorithm exclusion - from biometric authentication and video compression to generic sorting and tokenisation

The Annexure applies the three-step test from paragraph 4.5.3.1: identify the steps, decide whether they are abstract or enabled with technical specifics, and conclude. Examples...

Published
Updated
Reading time
7 min
Views
14
Questions
6 answered
  • Expert Reviewed
  • Medium Complexity
Topic
Patent
Published
October 4, 2026
Last updated
Oct 4, 2026
Reading time
7 min
0:00
Last updated: October 2026Verified against: Government sources

Examples 35 to 40 in Annexure I of the CRI Guidelines 2025 illustrate the algorithm limb of section 3(k). Four are claims tied to specific technical components and are found patentable; two are generic procedures, a sorting method and a text tokeniser, found excluded as pure algorithms.

The Office revises its guidelines, so check the current version on ipindia.gov.in. For the test itself see the algorithm exclusion article. A claim with a section 3(k) algorithm objection can be answered by a patent objection reply that matches it to the nearest example below.

What the Annexure is

Annexure I is described as a non-exhaustive indicative list of additional examples that illustrate allowable and non-allowable claims under the four limbs. The claims are invented by the Office and are not granted patents. This article summarises each in TaxClue's words.

The six examples

No.FieldOutcome in the AnnexureDeciding reason
35Fingerprint authentication on a mobile device: a neural network extracts a feature vector in an isolated secure processor, the vector is encrypted by a hardware engine and matched against encrypted templates to grant or deny accessNot excludedThe algorithm is realised through specific hardware and software components (sensor, secure enclave, hardware encryption) and answers the real problem of protecting biometric data
36Compressing surveillance camera video: a sensor captures frames, a hardware accelerator performs motion estimation, a standard codec is tuned for low latency, and the stream is sent wirelessly to a serverNot excludedEach step is tied to a technical component, and the claim solves low-latency, bandwidth-efficient streaming; the algorithm is not claimed in the abstract
37Noise suppression in a hearing aid: a microphone array, a digital signal processor running an adaptive filter, and an amplifier and speakerNot excludedThe adaptive filter is a real-time process interacting directly with hardware to solve background noise in a medical device
38Routing packets in a 5G network: a base station router uses live congestion metrics from edge nodes, applies a shortest-path method and updates routing tables as the network changesNot excludedThe claim is integrated with real network infrastructure and metrics and solves a technical network problem
39A generic sorting method: choose a pivot, partition, sort the parts recursively, combineExcludedA sequence of abstract steps with no technical implementation, integration or application to a real problem
40A text tokeniser: scan for delimiters, split, output the tokensExcludedThe same: abstract steps, no technical framework and no technical problem solved

What the passing examples have in common

Looking across 35 to 38, the Annexure's step 2 reasoning lists the same kinds of feature each time.

  • A physical or system component is named (a capacitive sensor and secure enclave; a camera sensor and hardware accelerator; a microphone array and signal processor; a base station router and edge nodes).
  • The algorithm runs on that component and interacts with it, so it is not presented as a stand-alone procedure.
  • The problem is technical: protecting biometric data, minimising delay over a limited wireless link, separating speech from noise, routing with low latency.
  • The claim goes beyond a label such as "apply an algorithm": it says what the algorithm takes, from where, and what happens to its output.

What the excluded examples lack

Examples 39 and 40 are each a short list of procedural steps. The Annexure notes that they do not say how the steps are realised in hardware, how they are integrated with a system, or what technical problem they solve. It gives illustrations of what could have been added, such as use in database performance in a cloud environment or in a speech recognition engine, to show the kind of context that is missing. The reasoning ends that each is a pure algorithm "without technical enablement or application".

The lesson is not that sorting or tokenising can never be part of a patent. It is that the claim, on its own wording, must carry the technical context. A sorting method claimed as the way a specific memory controller reduces access time is a different claim from the bare method.

Using the examples in a reply

ObjectionAnswer
"The claim is an algorithm"Identify the components in the claim that carry out each step, as in Examples 35 to 38
"The claim is an abstract sequence of steps"Add the technical context from the description: the device, the data source, the problem solved
"Standard algorithm used"A known algorithm can sit inside a patentable claim if its application is technical; point to the technical effect, as the Annexure does for shortest-path routing in Example 38

The paragraph 3.5.9 ruling in the Guidelines points the same way, that the inventive feature then lies in the implementation. See our article on the court rulings.

A worked example (invented)

Sangam Traffic files a claim to a method of finding the shortest route between two junctions using a graph search. The examiner relies on the pattern of Example 39. Sangam's description shows that the search runs on roadside controllers that receive congestion readings every few seconds and update signal timing. Sangam amends the claim to recite the controllers, the readings and the update of the timing, relying on the contrast between Example 39 and Example 38. The search method is unchanged, but the claim now ties it to equipment and a technical problem.

Common lapses

  • Claiming a standard algorithm "implemented on a computer" with nothing else.
  • Leaving the technical context in the description and out of the claim.
  • Describing a hardware component as a "module" with no structure.
  • Treating Examples 35 to 38 as a promise: they are illustrations of the test and every claim is judged on its own facts.

Need help with an algorithm objection?

If an examiner has called your claim an abstract algorithm, we can compare it with Examples 35 to 40 and identify which technical features need to be in the claim. See patent objection reply, and read next about the computer programme per se examples in the same Annexure.

Key takeaways

  • Examples 35 to 38 are not excluded because each ties the algorithm to named hardware and a real technical problem.
  • Examples 39 and 40 are excluded as pure algorithms with no technical enablement.
  • The deciding feature is what the claim says about implementation, not the algorithm used.
  • Known algorithms can appear in patentable claims.
  • The examples are indicative, and the Act and Rules prevail.

Read next

Disclaimer: Based on the manuals and guidelines published by the Office of the Controller General of Patents, Designs and Trade Marks that are named in the article, as consulted on 4 October 2026. They are guidance and do not have the force of law; the Patents Act, 1970 and the Patents Rules, 2003 as amended (including the 2024 amendment rules) prevail, and the current versions on ipindia.gov.in should be checked. This article is general information, not legal advice; check the official text before acting.

Quick recapKey facts & short answers

Key Facts About CRI

  • Applies in: All states across India, under the relevant central law.
  • Mode: Mostly online via the official government portal.
  • Typical timeline: Ranges from a few days to a few weeks depending on the case.
  • Non-compliance: May attract penalties, interest or late fees.
  • Expert help: TaxClue completes the entire process end to end for you.

Which algorithm examples pass?

Examples 35, 36, 37 and 38.

Which are excluded?

Examples 39 (generic sorting) and 40 (text tokenising).

Compliance is cheapest on the day it falls due and gets more expensive every day after.

— TaxClue Compliance Desk

CRI: a key compliance topic in Indian tax and corporate law that businesses and individuals must understand to remain compliant.

Related Services & Guides

Was this article helpful?
About the author
13,350 articles
Vikas Sharma Verified expert Tax & Compliance Expert

Experienced in company registration, GST, trademark, and compliance. Helping Indian businesses stay compliant.

Last reviewed: Live

Disclaimer: This article is for general informational purposes only and does not constitute professional tax, legal or financial advice. Laws, rates and due dates change and can vary by individual case — always verify with the relevant government source (e.g. mca.gov.in, incometax.gov.in) or consult a qualified professional before acting. TaxClue accepts no liability for decisions taken based on this content.

People also ask

Questions, answered

Short, direct answers to the 6 questions readers ask most on this topic.

Examples 35, 36, 37 and 38.

Examples 39 (generic sorting) and 40 (text tokenising).

Example 38 is integrated with a 5G network, uses live metrics and updates routing tables; Example 39 has no technical context.

Yes, if the claim shows the technical implementation and problem, as the passing examples do.

The Annexure excludes the bare tokeniser. It lists contexts such as a speech recognition engine as the kind of technical application whose absence led to the result.

They are indicative guidance. The Act and Rules prevail.