AI agents are usually imagined inside a browser: drafting an email, writing code, booking a meeting, or quietly eating your spreadsheet for lunch.
But some of the biggest opportunities for agents may be happening far away from a laptop screen — in trucks, warehouses, job sites, dispatch centers, maintenance shops, and everywhere else that keeps the physical economy moving.
In our latest episode of The Neuron: AI Explained, Corey Noles and Grant Harvey sit down with Samsara Cofounder and CTO John Bicket at Samsara Beyond 2026 in Las Vegas to talk about what AI looks like when the work involves vehicles, frontline workers, equipment, safety teams, and actual miles on the road.
Samsara has spent years connecting fleets and frontline operations through dash cams, telematics devices, asset trackers, and operational software. The company’s next bet is that this network can become an AI layer: one that sees what is happening, understands what needs attention, and helps teams take action before a small issue becomes an expensive one.
That is a different version of the agent story than “AI can write your meeting recap.” In physical operations, a missed signal can mean a breakdown, delayed shipment, safety incident, or a manager drowning in alerts. The useful agent is not the one with the flashiest demo. It is the one that helps a dispatcher, driver, or maintenance leader make the next decision faster and with more context.
Our favorite moments
- 1:13: John breaks down Samsara’s AI push, from Agent Studio and AI cameras to bird’s-eye operational visibility.
- 2:43: Why AI at the edge matters when the data comes from vehicles, cameras, and connected equipment rather than just documents and chat windows.
- 5:46: The leap from paper workflows to AI — and why some companies may skip parts of the traditional “digital transformation” playbook altogether.
- 7:53: How Agent Studio is meant to give operations teams a practical, permissioned on-ramp to building agents without turning everyone into an AI engineer.
- 15:12: What it looks like when AI can communicate with drivers, not just summarize data for people at headquarters.
- 17:09: Why physical-world agents have a different job than software-only agents: they have to handle messy, real-time environments where context changes fast.
- 28:16: The shift from detecting driving events after they happen to identifying patterns that could signal risk earlier.
- 30:43: The hard conversation around cameras, privacy, driver trust, and why safety tools fail when workers experience them as surveillance.
- 35:07: AI ride-alongs: using AI to scale the kind of coaching and context that traditionally depends on a human sitting in the passenger seat.
- 37:07: John’s vision for the next few years of operations: less hunting through dashboards, more real-time visibility and guidance.
AI agents need to leave the office
Samsara’s pitch is not that one magical agent will run a fleet. It is that operational teams already have the data they need — spread across vehicles, cameras, logs, policies, maintenance records, routes, and driver workflows — and AI can help turn that data into action.
That is the logic behind Agent Studio: teams can configure agents around their own policies, permissions, triggers, and operational needs. In practice, that could mean creating a maintenance digest, flagging a geofence event, assisting a driver with a company-policy question, or surfacing the few risks that deserve a manager’s attention first.
The key word is first. Physical operations do not need another dashboard full of beeps. They need systems that reduce the “where is it?” tax: Where is the truck? Where is the shipment? Which vehicle needs attention? Which safety issue is actually urgent? Who needs a follow-up before the next shift?
AI is most useful here when it shrinks the pile of work rather than adding another screen to monitor.
More automation raises the value for trust
Samsara’s customers are not starting from zero with AI. Many already rely on AI cameras, in-cab safety alerts, automated event detection, and dashboards that surface operational risks, and they've largely grown to appreciate the protections those provide. Agents extend that foundation by helping teams interpret the information faster, automate follow-up work, and make existing insights easier to act on.
But the more active these systems become, the more carefully companies need to manage the experience for frontline workers. Better visibility can improve safety and help teams respond faster. Keeping messages and actions focused on that value helps prevent concerns about such things being used to catch them doing something wrong.
That is why the conversation around AI ride-alongs and coaching matters. The strongest use case is giving drivers useful support, recognizing safer behavior, and helping human managers focus their attention where their experience can make the biggest difference.
The long-term adoption test for ai in all work is whether agents can build on tools workers already use without becoming another source of noise or scrutiny. They have to be useful to operations leaders, understandable to frontline teams, and designed so people know what the system is doing and why.
Samsara’s broader strategy is to turn its connected hardware and operational data into an intelligence layer. Agent Studio pushes that strategy beyond identifying events and surfacing insights toward automating what happens next.
Because agents are not just coming for inboxes. They're now arriving in the real world.
Watch and listen now: YouTube
Dive deeper
- Samsara’s Agent Studio and agentic-AI announcement — a closer look at the workflows, permissions, and operational use cases behind the conversation.
- Everything AI Samsara announced at Beyond 2026 — our deeper breakdown of the company’s AI cameras, maintenance tools, tracking labels, and agent strategy.
