#

Key takeaways

• AIR emerged from stealth after raising $50 million across two seed rounds.

• Its platform discovers agents in company environments and evaluates related tools.

• AIR says it can block interactions that fail its security criteria.

• The company says it filters out about 27% of the add-ons and skills it finds online.

• The source does not report local customers or deployments.

What AIR says it does

AIR is an AI security startup focused on vetting AI-agent tools and add-ons. According to the source, its platform discovers agents in company environments. It then evaluates their skills, plug-ins, MCP servers, and add-ons.

The company also says it can block interactions that do not meet its security criteria. That makes the product sound less like a general AI tool and more like a control layer around agent use. The source does not provide technical implementation details, so any deeper description would be an assumption.

The funding round

TechCrunch reported that AIR raised $50 million across two seed rounds. The source says one round brought in $10 million and was led by Sequoia. It also says a second round brought in $40 million and was led by Greenoaks.

This funding detail matters because it shows investor support for a security-first approach to AI agents. Still, the source does not explain how the capital will be used. It also does not state any product roadmap, hiring plan, or market expansion plan.

Why the filtering claim matters

AIR says it currently filters out about 27% of the add-ons and skills it finds online. That is the clearest operational signal in the source. It suggests the company is applying a screening layer before tools are allowed into use.

The number should be read carefully. It describes AIR's own filtering result, not a broader market benchmark. The source does not explain the criteria behind the filter, so readers should avoid treating the figure as a universal safety rate.

Governance and operational considerations

The story points to a simple governance lesson. If agents can discover and use external tools, then access control becomes part of AI security. The source also implies that plug-ins, MCP servers, and add-ons can create risk if they are not reviewed.

That does not mean every organization needs the same controls. It does mean teams should know which tools agents can reach, and which interactions should be blocked. The source supports that general principle, but it does not provide a formal framework or checklist.

Morocco relevance

The source reports no Morocco-specific customers, deployments, or partnerships. For readers anywhere, the conditional lesson is straightforward: if you adopt agents, review tool access and third-party components before broad rollout.

What the report does not say

The source does not name AIR's customers. It does not describe pricing, product tiers, or deployment models. It also does not give details on compliance claims, certifications, or incident history.

That absence matters. It means the report is about funding and product positioning, not a full security audit. Readers should treat the announcement as a company update, not as proof of complete protection.

Bottom line

AIR's announcement combines funding with a clear security message. The company wants to find agents, inspect their connected tools, and block unsafe interactions. Its reported 27% filtering figure adds a concrete sign of how it approaches that task.

For organizations evaluating agent adoption, the story is a reminder to govern tool access early. The source does not prove local relevance in Morocco, but the general lesson is portable. Security controls should be part of the agent conversation from the start.