Summary

• Primo has raised $8 million in financing led by Headline and Global Founders Capital.
• Its platform combines devices, SaaS access, identity, and support so agents can take operational actions rather than merely answer questions.
• Granting agents authority over company systems increases the importance of permissions, approvals, audit trails, and identity controls.

Paris-based Primo has raised $8 million to push autonomous AI further into corporate IT operations, where agents are being asked to change access rights, provision software, manage devices, and act directly on company systems rather than simply explain what an administrator should do.

The round was led by Headline and Global Founders Capital, with additional backing from Arthur Waller, co-founder and chief executive of Pennylane, and Romain Niccoli, co-founder and chief executive of Pigment. Current reporting says Primo serves more than 400 customers across ten countries.

The company plans to use the financing for product development, recruitment, and international expansion, with the UK, Germany, Benelux, and the Nordics among the markets it intends to deepen.

Primo combines device management, employee identities, SaaS applications, software access, procurement, and IT support, allowing agents to act against the same operational information that human administrators would otherwise move between several tools to manage.

Agents become operators rather than assistants

Corporate IT contains substantial work that is repetitive without being trivial. A new employee may need hardware, configuration policies, email and collaboration accounts, role-specific software, security controls, and access to internal systems, while a departing employee requires much of that process to be reversed quickly.

Primo’s own product material describes agents that can respond to HR events, provision or revoke application access, order hardware, identify devices falling outside policy, patch software, conduct access reviews, and resolve routine support requests. Organisations can require human approval before specified actions are completed.

Once an agent can change permissions or alter the state of a device, answer quality is no longer the only control point. Businesses also need to know what the system is authorised to do, where it must stop, and what evidence remains after each action.

Primo says its platform maintains detailed audit logs and lets customers define approval rules, while bringing employee, application, device, and policy information into a common model. That integration creates the product’s main advantage and concentrates its risk at the same time.

Fragmented IT creates the opening

Smaller and mid-sized organisations often use separate products for mobile-device management, SaaS administration, identity, purchasing, ticketing, compliance, and software licences, leaving administrators to reconcile information between them manually.

Agents can automate individual tasks inside that environment, but their usefulness is limited when each operation depends on a different system and source of truth. Consolidation therefore becomes part of the autonomy proposition because the agent needs reliable information about the employee, device, permitted applications, and relevant policy before making changes.

A similar direction is visible in larger industrial IT programmes. Metsä has folded agentic AI into a wider consolidation of industrial technology operations, treating autonomous systems as part of the operating architecture rather than another isolated productivity tool.

Autonomy raises the standard for controls

The labour-saving argument works only if automation does not create more supervision and remediation than it removes. An agent that grants privileged access incorrectly, removes a valid account, or applies an inappropriate security change can turn administrative efficiency into an operational incident.

Agents consequently need permissions and boundaries of their own rather than unrestricted administrative authority. Organisations also need to distinguish low-risk actions that can run unattended from consequential changes requiring approval.

The integrations themselves become sensitive infrastructure. A platform able to order equipment, alter access, inspect device state, and connect to HR systems becomes a valuable target because compromising it can provide a route across parts of the technology estate that were previously separate.

Primo’s financing gives it more capital to expand that model through Europe, but its longer-term test will be whether customers become comfortable delegating authority rather than merely adopting another AI interface.