Can my SSE control AI agents on laptops?

Only the part of their traffic that passes through it. Local actions such as file reads, shell commands and stdio MCP servers stay on the device. An on-device control covers that part.

Which SSE products does this site cover?

Zscaler Generative AI Security and AI Access Security from Palo Alto Networks, plus Lasso Security's proxy integrations. We could not read Netskope's AI pages for this edition, and Cato Networks' AI controls from its acquisition of Aim Security are not recorded yet.

Is a proxy the right first step for cloud workspaces?

Often, yes, because the traffic of virtual desktops and cloud development environments already runs through infrastructure you control. The finder adds weight to the network position when most work happens there.

CONTROL TYPE · NET

Network and SSE proxies for AI control: what they see and what they miss

Short answer

A network proxy or security service edge (SSE) inspects web traffic to AI services as it leaves the device, without installing anything AI-specific on the laptop. If you already route traffic through an SSE, it is a natural place to add AI rules. It cannot see what an AI agent does locally, such as reading a file or talking to an MCP server on the same machine.

By The Device Desk · Endpoint AI Finder · Published 2026-10-11

Desk research from public vendor pages read on 27 September and 11 October 2026. Profiles follow our published method; we did not run the products.

Vendor pages read September and October 2026 · updated 2026-10-13

What does network or SSE proxy do?

Many companies already send web traffic through a secure web gateway or an SSE service so they can apply policy to it. Adding AI rules there means deciding which AI services are allowed, and inspecting what is sent to them.

The advantage is reach without new software on the device. The traffic of every managed device, and sometimes of cloud workspaces, already passes through the proxy.

Three products in this set describe a network position. Zscaler Generative AI Security shows the input prompts users send to GenAI apps, allows or blocks apps, blocks sensitive data with DLP and can isolate the browser to disable cut, paste and download. AI Access Security from Palo Alto Networks, linked from its Prisma Access and SASE pages, covers visibility into shadow AI across 4,000+ GenAI apps, control and protection against data loss. Lasso Security describes proxy integrations for AI usage control and says it works with Palo Alto Networks or CrowdStrike. Netskope's AI pages could not be read for this edition.

When do you need it?

You already route web traffic through an SSE or secure web gateway.
Much of the work happens in cloud workspaces or virtual desktops whose traffic you control.
You want to allow or block AI services centrally before deciding on anything device-level.

What can it not see?

A proxy sees only traffic that passes through it. Coding agents act locally: they read files, run commands and start MCP servers on the same laptop, and the MCP specification recommends that servers meant to run locally use the stdio transport, which never crosses the network. Encrypted or pinned connections may also limit what the proxy can read, and traffic from devices outside your network path is invisible to it.

Which finder answers point to NET?

+1if the answer to "Which AI tools do employees use at work?" is AI chat and assistants in the browser
+1if the answer to "Where does the work happen?" is Mostly in cloud workspaces or virtual desktops
+1if the answer to "How sensitive is the data AI tools can reach?" is Regulated data, secrets or production access
+4if the answer to "Which of these do you already run?" is An SSE or secure web gateway
+1if the answer to "When an AI action looks risky, what do you want to happen?" is Block it

Try the finder

Which products describe network or SSE proxy?

3 of 19 products describe network or SSE proxy on the pages we read, 0 partly, and 16 are not in our sources.

Alphabetical. From each product's own pages.
Product	Status	What their pages say	Source
AI Access Security (Palo Alto Networks)	Described	Visibility into shadow AI across 4,000+ GenAI applications, control, and protection against sensitive data loss, delivered with Palo Alto Networks' SASE.	Source: Palo Alto Networks: AI Access Security · Read 11 Oct 2026
Akamai Workforce Protector	Not in our sources	Not described on the pages we read.	None
Bay	Not in our sources	Not described on the pages we read.	None
Bloom Security	Not in our sources	Not described on the pages we read.	None
Cloudflare AI Gateway	Not in our sources	Not described on the pages we read.	None
CrowdStrike Falcon Guardian	Not in our sources	Not described on the pages we read.	None
Ent	Not in our sources	Not described on the pages we read.	None
Glow Security	Not in our sources	Not described on the pages we read.	None
Harmonic Security	Not in our sources	Not described on the pages we read.	None
Koi (Palo Alto Networks)	Not in our sources	Not described on the pages we read.	None
Lasso Security	Described	Proxy integrations for AI usage control; the page says it works with Palo Alto Networks or CrowdStrike.	Source: Lasso Security: AI usage control · Read 27 Sep 2026
Neo	Not in our sources	Not described on the pages we read.	None
Noma Security	Not in our sources	Not described on the pages we read.	None
Onyx Security	Not in our sources	Not described on the pages we read.	None
Pluto Security	Not in our sources	Not described on the pages we read.	None
Prompt Security (SentinelOne)	Not in our sources	Not described on the pages we read.	None
Sun Security	Not in our sources	Not described on the pages we read.	None
Zenity	Not in our sources	Not described on the pages we read.	None
Zscaler Generative AI Security	Described	See all input prompts users send to apps, allow or block apps, block sensitive data with DLP across GenAI, and isolate the browser to disable cut, paste and download.	Source: Zscaler: Secure the use of generative AI · Read 11 Oct 2026

Not in our sources is not a no. Ask the vendor, and send us a public source if one exists.

What should you ask?

Questions for your SSE or proxy vendor:

Which AI services do you recognise, and how quickly do you add new ones?
Can you inspect prompts and uploads to AI services, or only allow and block domains?
Can you warn the user rather than only block?
What do you see of coding agents and MCP traffic, local and remote?
Do you read the MCP Mcp-Method and Mcp-Name headers that the 2026-07-28 specification added for intermediaries?

Sources

Source: MCP Security Best Practices (2026-07-28) · Read 27 Sep 2026

Source: Zscaler: Secure the use of generative AI · Read 11 Oct 2026

Source: Palo Alto Networks: AI Access Security · Read 11 Oct 2026

Source: Lasso Security: AI usage control · Read 27 Sep 2026

Learn more

AI controls for cloud workspaces and virtual desktops
When the work runs on a server, the endpoint that matters is the server, not the laptop.
Fitting AI controls next to the EDR, SSE and DLP you already run
The finder's fifth question asks what you already run. The answer often decides how a new control reaches your devices.