- 2023: Corma founded, launching AI copilot Cormie in 2026 to manage SaaS governance via plain language.
- Human-in-the-loop (HITL): Mandatory approval required for all AI actions to prevent unauthorized changes.
- Data integration: Connects to workplace tools via APIs, SSO, and browser extensions for accurate governance.
Experts would likely conclude that Corma's conversational AI approach represents a significant step toward democratizing SaaS governance while maintaining critical security and compliance safeguards.
The End of IT Tickets? Corma Launches Conversational SaaS Governance
PARIS – November 16, 2026 – The modern enterprise runs on a sprawling web of Software-as-a-Service applications, but the operational machinery required to govern this ecosystem remains stubbornly rooted in the past. For most organizations, requesting a new software license, auditing access rights, or simply figuring out why a departing employee still has an active CRM account requires a convoluted dance with the IT help desk. Today, a Paris-based startup is attempting to rewrite that operational playbook. Corma, a software license and access governance platform founded in 2023, has officially launched Cormie, an AI copilot designed to manage workplace software using plain language.
By allowing non-technical teams across human resources, security, and finance to query live identity data and initiate workflows via a conversational interface, the firm is targeting one of the most persistent operational bottlenecks in the 21st-century economy: the IT ticketing queue. The release highlights a growing shift in enterprise software toward conversational AI interfaces that bridge siloed departmental data, lowering the technical barrier for non-IT staff to manage access reviews and license costs while maintaining strict auditability.
Democratizing Software Governance
The traditional model of SaaS governance is inherently centralized, placing an immense operational burden on IT departments. When HR needs to onboard a new cohort of employees, or when finance wants to audit unused software licenses ahead of a vendor renewal, they are entirely dependent on IT administrators to manually pull reports across dozens of disparate systems. This dependency creates friction, slows down deployment, and obscures visibility into software spend.
The new conversational assistant aims to decentralize this capability. Built directly on top of the platform's centralized identity, access, and license data, the copilot allows users to bypass the help desk entirely for routine inquiries. An HR manager can simply type a request to identify which applications a new hire needs, while a finance director can ask the system to generate a chart of unused licenses costing the company money.
"An IT team should be able to roll out a new tool without worrying about how access will be managed. HR shouldn't have to chase down employee information to start onboarding, and finance shouldn't have to dig to see where the software budget is going," said Héloïse Rozès, CEO and Co-Founder of Corma.
"Cormie uses the data already in Corma to flag unused licenses before renewal, identify access that should have been removed, and help prepare onboarding once a hire is confirmed," she added. "Teams can then use it to carry out the next steps, with their approval required before any change is made."
The Guardrails of Delegated AI
While the operational benefits of conversational AI are clear, handing an autonomous agent the keys to enterprise identity and access management is a proposition fraught with cybersecurity risks. The integration of Large Language Models (LLMs) into critical infrastructure introduces novel threat vectors, most notably prompt injection and excessive agency.
Prompt injection, consistently ranked as the top risk by the Open Worldwide Application Security Project (OWASP) in its Top 10 for LLM Applications, involves manipulating an AI through malicious inputs. If an AI agent has the authority to provision software access, a cleverly disguised prompt hidden in an external document could theoretically trick the system into granting unauthorized administrative privileges. Similarly, excessive agency occurs when an AI is granted too much operational authority, allowing it to execute actions with unintended, widespread consequences.
To mitigate these enterprise security risks, the Paris-based firm has architected its copilot with strict, non-negotiable guardrails. Chief among these is a mandatory human-in-the-loop (HITL) requirement. The assistant cannot unilaterally execute write actions, such as provisioning a new account or revoking a license. Instead, it proposes the action, requiring explicit human confirmation before execution. Furthermore, every executed action is immutably recorded in a comprehensive audit log.
Crucially, the system enforces existing user role permissions on every data request. If a marketing coordinator asks the copilot for a list of executive salaries or administrative access to the company's cloud infrastructure, the system will block the request based on the user's established access control limits.
"You cannot deploy an AI agent into an identity governance environment without a hardcoded human circuit breaker," noted an enterprise security architect familiar with the evolving IGA landscape. "The moment you allow an LLM to autonomously modify access control lists without human verification, you are one hallucination away from a critical breach. The requirement for explicit approval, combined with strict adherence to existing role-based access, is the only way this technology passes a compliance audit."
The Battle for the Vertical Copilot
The launch of this new assistant places the startup squarely in the middle of a broader battle for the enterprise AI ecosystem. Tech giants like Microsoft and Google are aggressively pushing generalized copilots designed to assist with everything from drafting emails to summarizing meetings. However, a parallel trend is emerging: the rise of verticalized, domain-specific AI agents.
In the SaaS management and Identity Governance and Administration (IGA) space, incumbents and challengers alike are racing to integrate AI. Platforms such as SailPoint, BetterCloud, and Torii have all introduced various forms of AI-driven discovery, workflow automation, and conversational insights. What distinguishes specialized copilots from general-purpose enterprise AI is their deep integration into specific operational workflows and their ability to interface with complex, domain-specific data structures.
The newly announced tool leverages a proprietary assessment of data completeness and reliability when generating answers and alerts. By connecting to workplace tools with or without direct Application Programming Interfaces (APIs)—utilizing a mix of identity provider connections, single sign-on integrations, and browser extensions—the platform aims to build a highly accurate, normalized dataset. This focus on data hygiene is critical; an AI copilot is only as effective as the underlying data it relies upon, and hallucinations in the context of software licensing can lead to significant financial or security liabilities.
Automating the Future of Work
Beyond simple Q&A functionality, the true operational innovation lies in the platform's workflow module. Administrators can describe complex automations in plain language, instructing the AI to build reusable playbooks for tasks like cleaning up dormant accounts or orchestrating cross-departmental offboarding sequences. These playbooks can then be shared across the workspace, effectively democratizing the creation of IT automations.
The company has also signaled its intent to embed these capabilities directly into the communication channels where modern teams already operate. Plans are underway to bring the conversational assistant to Slack and Microsoft Teams, linking each user to their existing identity profile to ensure seamless, governed access from within the chat interface.
By transforming complex identity governance and SaaS management tasks into natural language conversations, the industry is witnessing a fundamental shift in how enterprise software is administered. The ability to distribute these operational responsibilities across departments, without sacrificing security or auditability, represents a critical evolution in the quest for organizational efficiency. As companies continue to grapple with software sprawl and the rising costs of digital infrastructure, the integration of specialized, governed AI agents may well become the standard operating procedure for the modern enterprise.
Topics & Related
Agentic AI
Software & SaaS
📝 This article is still being updated
Are you a relevant expert who could contribute your opinion or insights to this article? We'd love to hear from you. We will give you full credit for your contribution.
Contribute Your Expertise →