scripod.com
The Biggest AI Deployment Nobody Talks About | Samsara CEO Sanjit Biswas

Overview

Shownote

Highlights

Transcript

Chapters

Pins

The Biggest AI Deployment Nobody Talks About | Samsara CEO Sanjit Biswas

The MAD Podcast with Matt Turck

3 DAYS AGO
The MAD Podcast with Matt Turck

The MAD Podcast with Matt Turck

3 DAYS AGO
OverviewShownoteHighlightsTranscriptChaptersPins

Shownote

Sanjit Biswas runs what may be the largest AI deployment in the physical world — and almost nobody in AI talks about it. Samsara (NYSE: IOT), the ~$20B company he co-founded after selling Meraki to Cisco for $1.2B, puts AI on millions of trucks, cranes, an...

Highlights

In this episode, Matt Turck interviews Sanjit Biswas, CEO of Samsara, about the company's massive deployment of physical AI across millions of vehicles and industrial assets. They discuss Samsara's data scale, edge AI, AI agents, autonomous trucking, and the infrastructure demands of the AI boom.
00:00
Physical AI is transforming industrial operations at massive scale.
02:30
Fusing diverse data sources unlocks the true potential of physical AI.
03:04
AI shifts from reporting to actionable insights.
04:36
Physical industries represent 40-50% of global GDP.
06:07
Mistakes in physical operations can be life-threatening.
07:12
Samsara digitizes physical operations with AI and hardware.
08:22
Preventing 380,000 crashes and reducing CO2 emissions.
09:44
A quick glance at a phone can move a car the length of a football field.
11:28
A research project became a company.
13:42
Curiosity drove us into physical operations.
15:43
The agentic layer is the arms for action.
16:39
AI automates tasks that previously required human intervention.
18:36
BLE trackers last three to six-plus years
19:49
A disposable tag with a 45-day battery life.
21:08
A black box reveals engine and fuel dynamics.
22:16
AI dashcams provide real-time feedback to drivers.
23:28
Samsara's in-cab tech evolved from safety to AI communication.
24:49
Sensors collect data, stored in cloud, automated processes run on top.
26:32
Customers see ROI immediately without needing to be sold on AI.
27:52
Our moat is the proprietary data from the physical world.
29:22
Shared data improves safety and infrastructure for all.
31:23
Edge devices handle real-time inference for low-latency alerts.
32:35
We use a mix of AI models at the edge.
33:52
Generative AI enables automated video reasoning for driving events.
35:57
A hybrid model strategy balances frontier power with edge efficiency.
36:50
Agents autonomously execute multi-step tasks over longer time horizons.
37:56
Automating warranty checks cuts hours to under a minute.
38:50
Starting with low-risk AI use cases like warranty claims.
40:10
Combining workflow with data creates a significant unlock.
42:07
AI agents can get distracted or lost in loops.
43:13
Falling costs and rising capabilities will unlock new AI applications.
45:17
AI dash cams offer coaching but raise Big Brother concerns.
46:28
Dash cams primarily exonerate drivers in accidents.
48:48
Transparency and change management are key to gaining frontline trust.
50:37
Robots handle repetitive tasks, humans manage exceptions.
53:19
Customers are labor-limited, so they welcome automation.
54:51
Robotaxis will arrive faster due to uniform operations.
56:34
Data centers drive 90% of grid demand.
58:16
AI reduces mental load, letting workers focus on skilled tasks.
59:54
Samsara's work is economically important.

Chapters

Intro: The biggest AI deployment nobody talks about
00:00
What is physical AI?
01:16
From IoT dashboards to agentic action
03:04
Why physical AI is harder than software AI
04:36
Safety, cybersecurity, and real-world consequences
06:07
What Samsara does
07:11
$2B ARR, 25 trillion data points, and 380,000 crashes
08:22
How AI can prevent road accidents
09:44
From an MIT research project to Meraki
11:28
Learning physical operations from scratch
13:42
Samsara’s stack: sensors, intelligence, and action
15:42
Inside Samsara’s industrial asset trackers
16:39
Bluetooth, battery life, and connected infrastructure
18:36
A disposable tracking device built like a sticker
19:49
Vehicle gateways and engine diagnostics
21:08
How AI dash cams coach drivers in real time
22:16
Turning the dash cam into an AI interface
23:28
Organizing physical-world data in the cloud
24:49
Selling AI to traditional industries
26:32
Is Samsara’s real-world data its AI moat?
27:52
The network effects of covering 99% of U.S. roads
29:22
Edge AI versus cloud AI
31:23
The models running inside Samsara’s devices
32:35
Generative AI and video reasoning
33:52
Which AI models does Samsara use?
35:57
Inside Samsara Agent Studio
36:50
How an AI warranty agent works
37:56
Starting with practical, lower-risk automation
38:50
Combining agents, workflows, rules, and guardrails
40:10
What today’s AI agents still cannot do
42:07
AI ride-alongs and the future of driver coaching
43:07
Is workplace AI becoming Big Brother?
45:17
How cameras can protect and exonerate drivers
46:27
When AI becomes the judge of your work
48:48
Robots, humanoids, and mixed human-machine fleets
50:37
Samsara’s role in autonomous operations
53:19
How quickly will autonomous trucking arrive?
54:50
AI data centers and America’s infrastructure boom
56:32
Should lawyers become plumbers? Demand for tradespeople
58:16
Closing thoughts
59:54

Transcript

Sanjit Biswas: These are not the tokens you're going to find online, Like, you can't crawl Reddit and find out about what happened on a construction site. The Samsara system and a given day were driving 99% of the U.S. Roads, usually multiple times a day. ...