App. No. 19/629,591

obtain an input ... cause transmission of the input (claim 6); generate a cryptographic attestation of a routing decision ... maintain the cryptographic attestation in a distributed ledger (claim 7); store the input signature and each accessed agent signature in a persistent memory structure (claim 8)

For counsel to weigh: at Step 2B the examiner concludes the additional elements are 'well-understood, routine, and conventional,' supporting that conclusion only with MPEP § 2106.05(d)(II) citations and generic case categories ('receiving or transmitting data over a network,' 'storing and retrieving information in memory'). Under Berkheimer v. HP and MPEP § 2106.05(d)(I), a finding that an element is well-understood, routine, and conventional must be supported by one of the recognized evidentiary showings (a citation demonstrating the element is widely prevalent, an express applicant admission, a court holding on that element, or a § 1.132-type factual basis). Counsel may press that the office action supplies no such factual support for the more specific additional elements — notably claim 7's 'cryptographic attestation of a routing decision' 'maintain[ed] ... in a distributed ledger' and claim 8's 'persistent memory structure' — leaving the conventionality conclusion unsupported for those elements. This is an evidentiary defect counsel can raise without conceding the abstract-idea characterization.

• —Office action (claim 6, Step 2B): 'The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II)...)'
• —Office action (claim 7): treats 'maintain the cryptographic attestation in a distributed ledger' as extra-solution activity and 'generate a cryptographic attestation of a routing decision' as adding the words 'apply it,' without any evidentiary support for conventionality
• —Claim 7: 'generate a cryptographic attestation of a routing decision that identifies the at least one AI agent of the selected AI agent set; and maintain the cryptographic attestation in a distributed ledger.'

Risk The examiner may cure the gap on the next action by adding a Berkheimer citation or expressly designating the distributed-ledger/persistent-memory elements as notoriously well-known, so this argument may be temporary. It also does not by itself defeat the Step 2A abstract-idea finding, so counsel should pair it with a Prong Two practical-application position.

determining an input signature by applying one or more hash functions to the feature set (claim 6); perform a bitwise operation between the input signature and an agent signature to determine a distance value (claim 9); wherein the one or more hash functions comprise locality-sensitive hash functions (claim 12)

For counsel to weigh: the examiner's Step 2A Prong One characterization equates applying hash functions to the mental act of a human determining 'a distinct identifier for each and every AI agent' (office action, claim 6 Prong One analysis). That characterization is contestable under MPEP § 2106.04(a)(2)(III), which limits the mental-process grouping to concepts that can PRACTICALLY be performed in the human mind. Claim 9 requires 'perform a bitwise operation between the input signature and an agent signature to determine a distance value,' and claim 12 narrows the hash functions to 'locality-sensitive hash functions' — operations that a person cannot realistically carry out with pencil and paper on the feature-set data at issue. Counsel may argue the examiner recharacterized a specific computational operation as an abstract 'identifier' comparison and thereby understated what the claim actually requires, so the mental-process finding does not fit the limitations as claimed.

• —Office action (claim 6, Step 2A Prong One): 'For example, a human can determine a distinct identifier for each and every AI agent.'
• —Claim 9: 'perform a bitwise operation between the input signature and an agent signature to determine a distance value'
• —Claim 12: 'wherein the one or more hash functions comprise locality-sensitive hash functions'

Risk The examiner may reply that a hash is merely a rule that could be executed by hand for small inputs and that the claim recites the function at a high level of generality without tying it to a specific unmind-able computation. Prosecution-history caution: emphasizing that the operations require machine computation should avoid conceding that any hardware performing the operation is merely 'generic,' which could undercut a later Step 2B inventive-concept position.

Strategy check: re-ranked from #4 — For claim 20 the Berkheimer target (persistent storage) is expressly recited as conventional in the MPEP, so the Prong Two routing-path argument is the stronger claim-20-specific hook and should outrank the Step 2B storage argument here.

generate a cryptographic attestation of a routing decision that identifies the at least one AI agent of the selected AI agent set; and maintain the cryptographic attestation in a distributed ledger (claim 7)

For counsel to weigh: the office action treats claim 7's attestation generation as merely 'apply it' plus 'mere data output' and its distributed-ledger maintenance as extra-solution activity plus a generic computer component. Counsel may argue that generating a cryptographic attestation of the routing decision and committing it to a distributed ledger imposes a concrete, tamper-evident recordkeeping constraint on the routing output that goes beyond generally linking the exception to a field of use, and may weigh whether this qualifies as a meaningful additional element under MPEP § 2106.05(a)/(e) rather than insignificant post-solution output under § 2106.05(g). This is a Prong Two argument specific to the dependent claim that can survive even if the independent-claim Prong Two challenge is contested.

• —Office action (claim 7, Step 2A Prong Two): 'these limitations also amount to extra solution activity because it is a mere nominal or tangential addition to the claim, amounting to mere data output (see MPEP 2106.05(g)).'
• —Claim 7: 'maintain the cryptographic attestation in a distributed ledger.'

Risk The examiner will likely maintain that recording a result in a ledger is classic post-solution data output and that ledgers are recited generically. Prosecution-history caution: leaning on the ledger/attestation as the source of eligibility may effectively narrow the dependent claim's importance and could be read as conceding the independent claims lack a practical application.

compare the input signature to the accessed agent signatures to select one or more AI agents ... cause transmission of the input to at least one AI agent of the selected AI agent set (claim 6); comparing the input fingerprint to the accessed agent fingerprints to select one or more AI agents (claim 13)

For counsel to weigh: at Step 2A Prong Two the examiner treats the AI-agent, signature, and hash-function recitations as merely 'generally linking the use of the judicial exception to a particular environment of field of use' (office action, claim 6 Prong Two) and treats obtaining input and causing transmission as insignificant extra-solution activity. Counsel may argue instead that the claimed ordered combination — deriving a hash-based signature/fingerprint of a feature set and comparing it against agent signatures to route an input among a plurality of AI agents — is a specific technique for improving the operation of a multi-agent routing system rather than an abstract idea merely applied on a computer, in the vein of the improvement analysis at MPEP § 2106.05(a) (Enfish/McRO line). The locality-sensitive hashing (claim 12) and bitwise-distance selection (claim 9) supply a concrete, rule-based routing mechanism. This is a Prong Two path because, if the exception is integrated into a practical application, the analysis ends without reaching Step 2B.

• —Office action (claim 6, Step 2A Prong Two): 'The recitation of AI agents, input signature, and hash function amounts to generally linking the use of the judicial exception to a particular environment of field of use (See MPEP 2106.05(h)).'
• —Claim 6: 'compare the input signature to the accessed agent signatures to select one or more AI agents from the plurality of AI agents'
• —Claim 12: 'the one or more hash functions comprise locality-sensitive hash functions'

Risk The examiner will likely respond that the alleged improvement is to the abstract matching idea itself, not to computer technology, and that the specification must identify a concrete technical improvement. Counsel should confirm the as-filed specification actually describes a technical benefit (e.g., routing efficiency or accuracy) to anchor this argument. Prosecution-history caution: framing the invention around a particular technical improvement may narrow claim scope in the file wrapper.

wherein the feature set comprises one or more of: an entity associated with the input, an intent associated with the input, or a domain identifier associated with the input

For counsel to weigh: the office action analyzes claim 10 two different ways — first characterizing the feature-set limitation as 'extra-solution activity of gathering data' (pages 12-13), then alternatively as a mental process because 'a human can create a subset of data based on the original input set' (pages 13-15). Counsel may note that this shifting characterization reflects uncertainty about which judicial-exception grouping actually applies and may argue the limitation's 'entity/intent/domain identifier' content further specifies the feature set in a way that is not merely an abstract data-gathering step. This is a narrow point best used to reinforce the Prong One and practical-application arguments rather than as a standalone basis.

• —Office action (claim 10, first analysis): the feature-set limitation 'amounts to extra-solution activity of gathering data for use in the claimed process'
• —Office action (claim 10, second analysis): 'For example, a human can create a subset of data based on the original input set.'

Risk The examiner may respond that presenting alternative grounds is permissible and that both grounds independently support the rejection, so an inconsistency argument alone is weak. Best deployed only as support for the stronger Prong One / practical-application positions, not on its own.