Cohere Acquires cubic to Build Its Secure Coding Business
cubic lets an engineer's review comments become instructions for later reviews, giving a team's judgment a life beyond the pull request that first required it. Cohere has acquired the London AI code-review company and plans to put its team at the center of a new secure coding business, bringing that practical engineering relationship into a larger enterprise AI platform.
The companies announced the acquisition on November 13, 2026, without disclosing financial terms. cubic says its product will continue running and remain supported, while Cohere describes a broader role for its technology inside enterprise software development. For buyers, the interesting question is how a tool that learns the habits of an engineering team expands into institutions with their own rules about code, infrastructure and responsibility.
What Cohere acquired
cubic reviews proposed software changes, finds bugs and checks how those changes interact with a larger codebase. Its product also includes codebase scans and tools intended to make human review easier. That places the company close to an expensive decision: whether a piece of software is ready to become everybody else's problem.
Founder and CEO Paul Sanglé-Ferrière leads the team joining Cohere. Y Combinator's company profile lists cubic as a London company founded in 2025 and a member of its Spring 2025 batch. The founders' launch account describes engineers using more AI code-generation tools and finding that review and quality assurance demanded increasing attention. Faster creation had brought another part of the working day into focus.
The acquisition announcement from Cohere positions cubic's team as the foundation of its secure coding vertical. Cohere says it intends to integrate enterprise coding capabilities into its platform. The target's own announcement keeps the immediate customer commitment concrete: continued service, support and improvements.
Neither announcement supplies a purchase price, cash-and-stock breakdown or earnout schedule. The acquisition is described as completed, but the public announcement date should not be mistaken for a separately verified legal closing date. There is also no disclosed timetable for a combined product release, leaving customers with a clear continuity statement and a developing integration story.
Engineering standards become part of the product
cubic's learning feature makes the operational appeal easier to understand. The company says its reviewer can learn from senior engineers' past comments and from replies and reactions to its own feedback. Teams can also describe standards in plain English, turning a local convention into a check applied to subsequent work.
The design gives experienced engineers another way to influence code they may never personally review. That matters because a useful review often contains knowledge about how this particular team operates: which pattern it accepts, which assumption needs checking, or which business rule should survive a refactor. Repeating the same explanation across pull requests takes time that an engineering manager would rather spend on a new decision.
That knowledge also needs an owner. cubic says users can edit or delete learned instructions and keep rules in a repository, where changes go through the team's ordinary review process. A mistaken lesson can otherwise travel further than the original mistaken comment. Keeping a route for correction makes the learning relationship more credible than a product that simply promises to absorb expertise.
The commercial interpretation is straightforward: the accumulated standards can become part of the value of keeping a reviewer in the workflow. Each useful instruction gives the team a reason to consider continuity when it evaluates another tool. That is an inference about the product's economics, rather than a disclosed retention metric or an estimate of what Cohere paid for it.
Deployment is a purchasing decision
Cohere has built its enterprise AI proposition around organizations retaining control over data and infrastructure. Its company history begins in Toronto in 2019, with co-founders Aidan Gomez, Nick Frosst and Ivan Zhang. Aidan Gomez remains CEO, and the business now offers models and applications for enterprise work.
Buying a review product gives that proposition a specific place to operate. Source code can contain the logic a company depends on, and a reviewer needs enough access to understand what a proposed change will affect. The person responsible for adopting the tool therefore has to consider access, supported environments and the relationship between review advice and permission to merge.
Here the current documentation deserves close reading. Cohere's announcement describes on-premises and air-gapped deployment capabilities as part of its secure coding strategy. cubic's enterprise FAQ, as retrieved on November 16, supports GitHub and GitHub Enterprise Cloud while saying self-hosted GitHub Enterprise Server, GitLab and Bitbucket are not yet supported. Those statements do not establish that every deployment described in the acquisition strategy is already available through the existing cubic offering.
A prospective customer should ask which configuration is supported in its own environment and how the review rules will be managed there. Existing product documentation and a buyer's future platform plans serve different procurement purposes. That distinction gives engineering and security teams a useful starting point for the integration conversation, without inventing a release date or treating a broad strategy as a product specification.
The relationship Cohere now inherits
Cohere's expansion reaches beyond coding. The same announcement confirmed the completion of its AISLE acquisition, adding a separate cybersecurity team and business. For the cubic transaction, the more immediate operating task is to carry an established reviewer and its users into a broader platform while preserving what made it useful to engineering teams.
The purchase also sits alongside Cohere's recently announced Series E initial close. The firm's October 29 funding statement describes a round targeting up to C$7B, which is a target rather than a disclosed C$7B already raised. The context is expansion toward a larger enterprise offering; the public record does not allocate that financing to a particular acquisition price or coding budget.
DevCuration's coverage of Hypercubic's mainframe work and IBM's AuthMind acquisition follows adjacent questions about existing software and the permissions around AI. These are separate businesses and transactions. Together, the subjects give operators a reason to examine how new AI capabilities meet systems and controls already carrying institutional obligations.
cubic's existing customers can follow the integration through a more familiar unit of progress: the next useful review. A senior engineer can inspect what the tool learned, correct a rule and see how it handles another change. Cohere now owns a company whose relationship with that engineer is renewed every time the reviewer has something worth saying.
Frequently Asked Questions
What happens to existing cubic customers after the acquisition?
cubic says its product will continue running, remain supported and receive improvements as its team joins Cohere. The announcements do not provide a combined-product release timetable or a new pricing policy.
How does cubic use a software team’s review standards?
cubic says it learns from senior engineers’ past review comments and from feedback on its own comments. Teams can edit or delete learned instructions and maintain review rules in version control.
Is self-hosted cubic available for every enterprise environment?
The acquisition announcement describes on-premises and air-gapped capabilities as part of Cohere’s secure coding strategy. cubic’s enterprise FAQ, retrieved November 16, 2026, supports GitHub Enterprise Cloud but says self-hosted GitHub Enterprise Server, GitLab and Bitbucket are not yet supported; buyers should verify the exact available configuration.
Which acquisition terms have been disclosed?
Cohere and cubic announced the acquisition on November 13, 2026, but did not disclose a purchase price, consideration breakdown or earnout schedule. The public announcement date is not a separately verified legal closing date.
Why does code review fit Cohere’s enterprise AI strategy?
Reviewing software changes brings AI into a workflow where engineering standards, source-code access and deployment controls matter to the buyer. Cohere says cubic’s team will establish its secure coding vertical and support integration of enterprise coding capabilities into its platform.
Where the Money Moved
The intelligence briefing of the innovation economy. Funding, M&A, debt and fund closes, read as market signal rather than deal announcements.
Subscribe to Where the Money Moved