Sanas.AI Inc. v. Krisp Technologies, Inc.

CourtDistrict Court, N.D. California
DecidedDecember 1, 2025
Docket3:25-cv-05666
StatusUnknown

This text of Sanas.AI Inc. v. Krisp Technologies, Inc. (Sanas.AI Inc. v. Krisp Technologies, Inc.) is published on Counsel Stack Legal Research, covering District Court, N.D. California primary law. Counsel Stack provides free access to over 12 million legal documents including statutes, case law, regulations, and constitutions.

Bluebook
Sanas.AI Inc. v. Krisp Technologies, Inc., (N.D. Cal. 2025).

Opinion

1 2 3 4 5 6 7 UNITED STATES DISTRICT COURT 8 NORTHERN DISTRICT OF CALIFORNIA 9 SANAS.AI INC., 10 Case No. 25-cv-05666-RS Plaintiff, 11 v. ORDER DENYING KRISP’S MOTION 12 TO DISMISS SANAS’ CLAIMS AND KRISP TECHNOLOGIES, INC., GRANTING IN PART AND DENYING 13 IN PART SANAS’ MOTION TO Defendant. DISMISS KRISP’S COUNTERCLAIMS 14

15 I. INTRODUCTION 16 Defendant and Counterclaim Plaintiff Krisp Technologies (“Krisp”) and Plaintiff and 17 Counterclaim Defendant Sanas AI (“Sanas”), two technology competitors and almost- 18 collaborators, move to dismiss each other’s claims and counterclaims, respectively. Sanas has 19 adequately pled trade secret misappropriation, co-inventorship and co-ownership, and false 20 advertising claims against Krisp. Accordingly, Krisp’s motion to dismiss as to these claims is 21 denied as set forth below. 22 Sanas’ motion as to Krisp’s state-law counterclaims is denied in part as well because Krisp 23 has adequately pled violations of Cal. Bus. & Prof. Code § 17043 and §§ 17200 et. seq. However, 24 since allegations of violations of Cal. Bus. & Prof. Code § 17044 cannot survive when based on a 25 defendant freely giving away a product, Sanas’ motion to dismiss is granted as to count three of 26 Krisp’s counterclaims as set forth below. 27 1 II. BACKGROUND1 2 Sanas and Krisp are competitors in the market for voice communication software. 3 Specifically, their technology improves phone and video call conversations in real-time by 4 converting accents and reducing background noise. It is most commonly deployed in call centers. 5 Krisp entered the market first with background noise suppression software. Concurrent 6 with the release of its noise suppression technology, Krisp filed a provisional patent application 7 that was awarded on August 24, 2021. 8 Sanas AI was founded a few years later by Maxim Serebryakov, Shawn Zhang, and 9 Andrés Pérez Soderi, three former Stanford classmates, along with Sharath Narayana, an 10 entrepreneur. The Sanas co-founders were originally focused on accent conversion. Between 2020 11 and 2021, they built out a database of accented human speech, including identifying and recording 12 “ideal” target accents. They then developed software that translates accents between parties in 13 real-time, fast enough and at a high enough quality that normal conversation can be carried out, 14 leveraging unique student-teacher machine learning architecture and parallel data generation. 15 They launched their first product in 2021 and filed a provisional patent application titled 16 Real-Time Accent Conversion Model in May 2021. Other applications followed, including six 17 patents by September 2025. 18 A. Collaboration Discussions Between Sanas and Krisp 19 Krisp alleges that by at least July 2021, it also was developing accent conversion 20 technology using student-teacher machine learning architecture. That fall, Krisp reached out to 21 Sanas about collaborating. Between October 2021 and November 2022, the parties explored 22 collaboration, and Sanas provided, upon Krisp’s solicitation, technical details about Sanas’ accent conversion technology. That technical information included performance details in the case of 23 strong accents, different dialects, and background voices; intelligibility of various converted 24 audio; end to end latency metrics when used with different platforms; support for different 25

26 1 For the purposes of these motions, unless otherwise stated, this order accepts well-pled factual 27 allegations made in the complaint and counterclaims as true. 1 microphone types; CPU utilization; details about the technology stack and current and future 2 technical specifications; and decisions and plans to build versus buy certain algorithms and 3 features. 4 Upon Sanas’ request, the parties entered a non-disclosure agreement (NDA) on November 5 17, 2021. The NDA acknowledged that Sanas “ha[d] disclosed, and may further disclose certain 6 confidential technical and business information. . . that [Sanas] desires [Krisp] to treat as 7 confidential.” The NDA defined “Confidential Information” as (i) written information that is 8 “marked ‘Confidential,’ ‘Proprietary’ or in some other manner to indicate its confidential nature,” 9 or (ii) oral information that is “reduced to a written summary [] that is marked in a manner to 10 indicate its confidential nature and delivered to the Recipient within thirty (30) days after its initial 11 disclosure.” Dkt. 39-2, Ex. A, § 2.A. Pursuant to the NDA, Krisp promised not to use Sanas’ 12 confidential information “for any purpose except to evaluate and engage in discussions” relating to 13 the licensing opportunity Krisp had solicited. Id., § 3. The NDA further provided that “[n]othing 14 in this Agreement is intended to . . . grant [Krisp] any rights in or to the Confidential Information 15 of [Sanas] except as expressly set forth in this Agreement.” Id., § 8. 16 The parties continued to explore collaboration, creating a Slack channel for exchange of information directly between the companies’ technical teams. At various times, Sanas also sought 17 an exclusivity agreement to enhance protection of the information it was sharing. Krisp refused. 18 Sanas proceeded nonetheless, expressly indicating to Krisp that it understood the NDA to provide 19 sufficient protection for the technical discussions the two companies were having. At the end of 20 October 2022, however, Krisp stopped responding to outreach from Sanas’ senior management. 21 On November 4, 2022, Krisp emailed Sanas stating that Krisp did not want to proceed with the 22 partnership for various reasons, including that Krisp was not yet interested in pursuing accent 23 translation. 24 B. Krisp’s Accent Conversion Technologies and Patent Filings 25 On April 27, 2023, sixteen months after Krisp first reached out to Sanas about licensing its 26 technology and five months after discussions terminated, Krisp publicly announced “early access” 27 1 to its own proprietary AI accent conversion technology. It posted an article on its company blog 2 announcing “Krisp AI Accent Conversion: Get Ready for a Communication Revolution.” The 3 article explains that Krisp “ha[d] been working tirelessly to create a technology that utilizes real- 4 time inflection changes to help customers understand agents better by dynamically changing agents’ accents into the customer’s natively understood accent.” It announced a range of dialects 5 for which the product already worked. 6 On July 21, 2023, Krisp filed two provisional U.S. patent applications, which issued as 7 U.S. Patent No. 12,205,609 (“the ’609 Patent”) and U.S. Patent No. 12,223,979 (“the ’979 Patent” 8 and, together with the ’609 Patent, the “Krisp Accent Conversion Patents”). The Krisp Patents cite 9 Sanas’ accent conversion technology patents and patent applications. 10 Sanas alleges that the Krisp Accent Conversion Patents do not disclose or claim any novel 11 or non-obvious approaches to generating parallel data for speech conversion but rather claim, as 12 Krisp’s own inventions, techniques and approaches discussed in Sanas’ earlier patent applications 13 and issued patents. Sanas also alleges that the Krisp Patents rely and build upon information 14 provided by Sanas employees, including co-founder Serebryakov, such that they should have been 15 named as joint inventors. Specifically, they assert that information regarding the process of 16 training the student machine learning algorithm with the teacher algorithm, which Sanas 17 employees shared with Krisp, are reflected in the ‘609 and ‘979 Patents.

Free access — add to your briefcase to read the full text and ask questions with AI

Related

Kewanee Oil Co. v. Bicron Corp.
416 U.S. 470 (Supreme Court, 1974)
Bell Atlantic Corp. v. Twombly
550 U.S. 544 (Supreme Court, 2007)
Ashcroft v. Iqbal
556 U.S. 662 (Supreme Court, 2009)
Skydive Arizona, Inc. v. Quattrocchi
673 F.3d 1105 (Ninth Circuit, 2012)
Dumas v. Kipp
90 F.3d 386 (Ninth Circuit, 1996)
Umg Recordings, Inc. v. Shelter Capital Partners Llc
718 F.3d 1006 (Ninth Circuit, 2013)
Zucco Partners, LLC v. Digimarc Corp.
552 F.3d 981 (Ninth Circuit, 2009)
Cel-Tech Communications, Inc. v. Los Angeles Cellular Telephone Co.
973 P.2d 527 (California Supreme Court, 1999)
In Re Tobacco II Cases
207 P.3d 20 (California Supreme Court, 2009)
Co-Opportunities, Inc. v. National Broadcasting Co.
510 F. Supp. 43 (N.D. California, 1981)
Diodes, Inc. v. Franzen
260 Cal. App. 2d 244 (California Court of Appeal, 1968)
Fisherman's Wharf Bay Cruise Corp. v. Superior Court
7 Cal. Rptr. 3d 628 (California Court of Appeal, 2003)
Pellerin v. Honeywell International, Inc.
877 F. Supp. 2d 983 (S.D. California, 2012)

Cite This Page — Counsel Stack

Bluebook (online)
Sanas.AI Inc. v. Krisp Technologies, Inc., Counsel Stack Legal Research, https://law.counselstack.com/opinion/sanasai-inc-v-krisp-technologies-inc-cand-2025.