Autonomous AI Agents for IT Teams: Primo Raises €6.9M
Paris-based startup Primo raised €6.9 million to build autonomous AI agents that diagnose and resolve IT operations issues without constant human oversight.
Primo, a Paris-based startup, has raised €6.9 million to build autonomous AI agents that handle IT operations tasks for enterprise technology teams. The funding will support development and deployment of agents designed to take on work traditionally handled by IT staff, such as diagnosing and resolving technical issues without constant human oversight.
The round positions Primo among a growing wave of European startups betting that AI agents — software capable of executing multi-step tasks autonomously rather than simply answering prompts — are ready for real deployment inside corporate IT departments.
Autonomous IT agents represent a shift from AI as a support tool to AI as an operator. Instead of an engineer querying a chatbot for troubleshooting steps, an agent can execute the diagnosis and remediation itself, changing how IT teams allocate time, skills and headcount. For organizations under pressure to modernize infrastructure while controlling costs, this is a concrete test case for how much operational work can be safely delegated to AI.
The bet also matters for the broader enterprise AI market: agentic systems that act rather than advise are the next frontier beyond generative chat tools, and IT operations — repetitive, rules-based, high-volume — is a logical proving ground before agents move into more customer-facing or judgment-heavy domains.
— €6.9 million raised by Primo to develop and deploy its autonomous AI agents for IT teams.
The headline is the funding, but the real story is what "autonomous" is being asked to prove. IT operations is an attractive testbed precisely because failures there are largely invisible to end customers — which makes it a low-risk place to learn hard lessons before agentic AI touches anything customer-facing.
Every "autonomous agent" pitch quietly depends on someone deciding what the agent is allowed to break. The winners here won't be the teams with the smartest models — they'll be the ones who design the escalation paths, audit trails and failure modes before the agent ever touches production. Treat this as a service-design problem wearing an engineering costume: the trust budget for AI acting without a human in the loop is earned in IT tickets long before it's spent on customers.