Radio
Now Playing
Quickyla Radio โ€” Click to play
Open โ†’
3 min left
Back to News

Brex monitors AI agents' network activity to manage risk

Brex treats AI agents as virtual employees with network access, monitoring their actions (not code) to manage risks. This approach aims to improve security and speed up AI adoption in enterprises by โ€ฆ

Brex assumes its AI agents could do anything โ€” so it watches the network, not the code
VentureBeat โ€” 10 August 2026
Text:
43 0 0

Brexโ€™s CEO Pedro Franceschi has laid out a new approach to safely running AI agents inside big companies. At the VB Transform 2026 conference, he argued that the industry should stop calling them โ€œagentsโ€ and start thinking of them as virtual employees with Slack logins, email addresses and meeting invites. The idea is to treat each AI like a new hire that can sit in on calls, read documents and answer internal messagesโ€”without the usual code-level scrutiny.

Franceschi says most companies still police AI by scanning its code, but that misses the point once the system starts acting on live data. Brexโ€™s solution is to watch what the AI actually does on the network instead of what it could do on paper. By giving OpenClaw, an open-source AI framework, real internal roles and monitoring its every move, Brex found it could keep the benefits of automation without the usual security headaches.

The company now runs OpenClaw in a handful of pilot roles, watching for unusual file access or messages that fall outside its assigned duties. Early results show fewer false alarms than traditional code scans and faster onboarding of new AI helpers. Franceschi insists the shift from โ€œagentโ€ to โ€œvirtual employeeโ€ is more than semanticsโ€”it forces clearer rules about what each AI is allowed to touch and when.

If the pilot scales, Brex plans to publish its security playbook so other firms can adopt the same network-first approach. The stakes are high: companies that treat every AI like a new colleague will move faster, but only if they can prove to regulators and auditors that the bots are behaving themselves. Franceschi reckons the next wave of enterprise AI will succeed or fail on trust, and monitoring the networkโ€”not the codeโ€”is how you earn it.

Read Full Story at VentureBeat โ†’
Verified Source
Advertisement
React:
Sources
Sponsored

More to Read

Apple AirPods Pro 3 tops 2026 wireless earbud rankings
๐Ÿ’ป Technology
Apple AirPods Pro 3 tops 2026 wireless earbud rankings
Wired ยท 10 days ago
The AI dictation app everyone is talking about just got a pโ€ฆ
๐Ÿ’ป Technology
The AI dictation app everyone is talking about just got a powerful new Notetaker
Android Authority ยท 15 days ago
Reddit is letting AI decide when your post breaks the rules
๐Ÿ’ป Technology
Reddit is letting AI decide when your post breaks the rules
Android Authority ยท 14 days ago
Iran voids 60-day nuclear negotiation deadline with US
๐ŸŒ World News
Iran voids 60-day nuclear negotiation deadline with US
France 24 ยท 2 days ago
Sanguinetti directs poetic debut on childhood in Argentina
๐ŸŽฌ Entertainment
Sanguinetti directs poetic debut on childhood in Argentina
Variety ยท 10 days ago
Idlib residents celebrate court's death sentence for Assad,โ€ฆ
โš”๏ธ War & Conflict
Idlib residents celebrate court's death sentence for Assad, former officials
Al Jazeera ยท 8 days ago
Full view