Industrial machines have had a big red button for over a century. The reason is simple: when a system can act in the physical world, an operator needs a way to stop it instantly — no menus, no configuration, no waiting. As AI agents move from answering questions to taking real actions across email, calendars, finance systems, and procurement portals, they need the same thing. Today ApiSpi is adding it: a one-click Emergency Stop.
Why Agents Changed the Calculus
A chatbot that gives a bad answer is an inconvenience. An agent that can send emails, post messages, create invoices, or submit forms is a different category of risk. As we argued in From Chatbot to Agent, the leap in capability is exactly what makes agents valuable — and exactly why control has to keep pace. The moment software can take consequential, hard-to-reverse actions on your behalf, "turn it off right now" stops being a nice-to-have.
What Emergency Stop Does
The button lives at the top of your profile and your dashboard, and it does one thing decisively: it cuts all connector and tool access for your account in a single tap.
- Instant — one click disables every connected service, so Aria can no longer reach email, calendars, AI providers, or any other tool.
- Non-destructive — your connectors, keys, and governance settings are untouched. Nothing is deleted; access is simply suspended.
- Reversible — the same button turns green and reads "Resume" so you can restore access the moment you're ready.
Visible Everywhere It Matters
A safety control that's easy to forget isn't much of a control. When Emergency Stop is active, ApiSpi makes it impossible to miss — a clear banner appears on your dashboard and at the top of the Aria chat, and even basic-mode replies carry a notice explaining that connectors are off and how to bring them back. No silent failures, no wondering why an agent suddenly can't reach a tool.
Part of a Layered Model
Emergency Stop isn't a replacement for fine-grained governance — it's the blunt instrument that sits above it. Day to day, the Policy as Code layer and the platform's approval gates handle the nuanced decisions about what an agent may and may not do. Emergency Stop is the override for the moments when nuance can wait and you just need everything to halt.
- An anomaly you're investigating — pause all activity while you check whether an agent behaved as expected.
- A credential or connector you suspect is compromised — sever access first, sort out the detail second.
- Change-freeze windows — keep agents read-only during a sensitive period without unpicking your configuration.
Control You Can Reach in One Tap
Trust in autonomous systems is built on the confidence that you can stop them. Emergency Stop is available now for every ApiSpi user — a small button doing an important job, and one more reason capability and control move forward together on the platform.