What a pasted prompt can expose
A fictional bank tests the data boundary before access widens.
Coming
See every demonstrationIBOSS AND MICROSOFT PURVIEW
Find where AI workflows expose data across iboss and Purview, with tests.
You run iboss at the network edge and Microsoft Purview in the tenant, and nobody can say which of the two catches a given AI data path, or which paths neither sees. In three weeks we confirm the boundary and the owners, review the selected policies and map who is responsible for what, run agreed synthetic tests in a non-production environment, and hand over a remediation roadmap with every untested path named.
01 Source02 Boundary03 Human decision04 Evidence
HOW IT WORKS
The control path in practice.
- Week one: confirm the boundary, licensing, owners, approved access, the AI workflows and the criteria; you approve the inventory and the validation plan.
- Week two: review the selected policies, map control responsibilities between iboss and Purview, and run the agreed synthetic tests where supported, with production read-only.
- Week three: reconcile the findings with the owners, present the roadmap, and record acceptance or the unresolved exceptions.
- Every recommendation ties to an approved source, a supported configuration or an observed test, and the unsupported paths are listed as such.
THE PROBLEM
Why do two control planes leave a gap?
Because each one assumes the other covers the path it does not see. The network edge sees the browser session; the tenant sees the labeled file; the AI workflow that moves a client record from one to the other is observed by neither until someone writes the responsibility map and tests it.
The assessment reviews the supported integration between iboss AI-powered SASE for Microsoft Purview and your Purview capabilities, with Entra ID for identity context, and separates configuration evidence from untested assumptions in every finding.
THE MICROSOFT INSIGHT
What does Purview enforce on the tenant side?
A Microsoft Purview DLP policy with the Microsoft 365 Copilot and Copilot Chat location stops Copilot and Cowork from processing files and emails that carry chosen sensitivity labels, which the responsibility map records as the tenant-side control for labeled content, with the network-side control for everything else.
A Microsoft Purview DLP policy with the Microsoft 365 Copilot and Copilot Chat location stops Copilot and Cowork from processing files and emails that carry chosen sensitivity labels; the item can still appear as a citation. This needs Microsoft 365 E5 or the Purview suite. (Microsoft Learn, checked Oct 31, 2026)
Write which plane owns each path and a synthetic test proves it. Assume both do and the untested path is the one the audit finds.
THE METHOD
What happens in the three weeks?
Boundary and plan, review and tests, reconciliation and roadmap.
- Week 1: confirm boundary, licensing, owners, approved access, AI workflows and criteria; you approve the inventory and the validation plan.
- Week 2: review the selected policies, map control responsibilities, run agreed synthetic tests where supported; production read-only, active tests in an approved non-production environment.
- Week 3: reconcile findings with owners, present the roadmap, record acceptance or unresolved exceptions, and identify prerequisites for a separate implementation.
THE EDGE
What does the network layer add that the tenant cannot see?
The browser session to an AI site the tenant does not license: Microsoft Entra Internet Access, for comparison, includes prompt injection protection, shadow AI discovery and web DLP in its Internet Access profile, and the responsibility map records which of those functions iboss performs in your estate and which gaps remain.
Microsoft Entra Internet Access, in the Microsoft Entra Suite or standalone with Entra ID P1 or P2, includes prompt injection protection, shadow AI discovery, TLS inspection and data loss prevention for web traffic, and its network controls for agents need a Microsoft Agent 365 license. (Microsoft Learn, checked Nov 4, 2026)
Map the edge functions and the tenant functions side by side and the gaps are rows with owners. Keep them in two teams and the gap is a meeting.
WHAT IT IS NOT
Does the assessment change policies or clean up content?
No. Production policy changes, implementation, software resale, license charges, content cleanup, incident response, custom connectors, compliance certification and any promise of inspection coverage are excluded; Azure Government and other sovereign clouds need separate validation.
Scope is one Microsoft 365 tenant and one iboss environment, up to five AI workflows, up to twenty-five existing policies and up to twenty synthetic validation cases, delivered remotely after access, licensing, owners and the signed scope are confirmed. You authorize Microsoft 365 and iboss permissions separately and remove them at closure; Azure Lighthouse administers neither and is not used as a delegation mechanism here. Raw production content is not needed and evidence stays in your systems.
AFTER TWO WEEKS
What will be working after two weeks?
- You hold the AI data-flow inventory and the scope record you approved in week one.
- You hold the control responsibility and coverage map, including the unverified and unsupported paths.
- You hold the policy and exception register with evidence references, severity, owners and proposed actions, and the synthetic validation results separating configuration evidence from untested assumptions.
- You hold the prioritized remediation roadmap and the acceptance record.
The work is delivered remotely inside your own Microsoft tenant, wherever your business is, and on site only when the work needs it.
HONEST ABOUT LIMITS
What this does not do yet
It does not block a typed prompt in Copilot with a generally available control today. Blocking a Copilot or Cowork response when the typed prompt contains chosen sensitive information types is in preview and rolling out to tenants. (Microsoft Learn, checked Oct 31, 2026, preview) Blocking a Copilot or Cowork response when the typed prompt contains chosen sensitive information types is in preview and rolling out; the responsibility map records the prompt path as preview on the tenant side and tests the network side where iboss covers it.
It does not change a production policy or clean up content; the roadmap names the implementation as separate work.
It does not promise inspection coverage; the unsupported paths are listed as unsupported.
THE ONE THING TO REMEMBER
Two control planes, one responsibility map, every path tested or named untested.
Said another way: boundary, map, synthetic test, roadmap.
Said a third way: a gap you can name has an owner; a gap you assumed away has a date.
SHOW YOUR WORK
See the control path on a fictional bank.
DIRECT ANSWERS
Questions before you decide.
What do we need in place?
An existing iboss environment and Microsoft Purview with the licenses the tests use, named security and data owners, authorized read access to the selected policy configuration, and a non-production environment for the active tests.
Does it touch production?
Production discovery is read-only. Active tests run only in a non-production environment you approve, on synthetic cases, and raw production content is never needed.
How much is in scope?
One Microsoft 365 tenant and one iboss environment, up to five AI workflows, up to twenty-five existing policies and up to twenty synthetic validation cases, over three weeks of remote delivery.
How is it bought?
Through a private offer requested from Simplicity IT and, once the listing is live, purchased through Microsoft Marketplace. Licenses and implementation are separate. No price appears on this site.
NEXT STEP
Start with a boundary you can explain.
Choose a focused check, or speak with the team about the workflow you need to improve.