BlueAspen
Data Security for AI · Agents + Employees
Data security for AI
Control what your AI reaches —
and what it brings back.
Across the agents you run and the employees using AI. BlueAspen inspects and controls the real data on every AI path — inline, at the moment of the call, not in a report after the fact — with validated DLP, redaction in place, and fail-closed enforcement on the paths that guard your data. The sensitive part is often in what comes back; that's the half we screen.
MCP tool · ehr.getPatient
screened
Request →
tools/call · { "patientId": "P-48213" }
← Result
{ "name": "Dana Whitfield", "dob": "1984-03-11", "ssn": [REDACTED], "mrn": [REDACTED], "plan": "PPO-Gold" }
Findings
PHI · SSN · 0.98
PHI · MRN · 0.95
Result leg · redacted in place · reply still valid JSON
Redact
Two surfaces, one engine
The agents you run
Agent → tool, agent → agent, agent → model
An agent calls MCP tools, other agents (A2A) and LLMs on its own — plus the MCP servers you expose inbound. The Agent Proxy and MCP Gateway inspect and control the data on every one of those paths.
The employees using AI
People → ChatGPT, Claude, Gemini & ~6,000 AI apps
Regulated data lands in prompts and uploads across the AI tools your people use on their endpoints — most of which security never approved. The Employee Proxy (forward proxy + endpoint agent) sees it.
One engine
One policy engine and one detection stack behind both surfaces — the same validated DLP and the same audit evidence whether the actor is an autonomous agent or a person. Posture differs by design: ingress and agent→tool paths fail closed; the employee egress proxy fails open, so no one is ever blocked from working.
Where we are
3
enforcement modes shippedMCP Gateway · Agent Proxy · Employee Proxy
22/22
outbound live end-to-endA2A + AWS Bedrock
2,900+
automated testsinbound passes the official MCP client SDK
60+
validated detection patternsPCI · HIPAA · PII · secrets
Why BlueAspen — depth, not a PII checkbox
Data, not traffic
i.
- We inspect and redact what AI brings back — on the result leg, where the leak happens
- Redaction applied in place, so the reply stays valid and the AI keeps working
Audit-grade
ii.
- Validated detection — Luhn, NPI, IBAN, Shannon entropy — across 60+ context-scored patterns
- A reason on every finding, tuned for low false-block. A compliance artifact, not a guess
Fail-closed
iii.
- On ingress and agent→tool paths, a call we can't judge never reaches the data
- A misconfigured gateway refuses to start (employee egress fails open by design)
Find the undeclared
iv.
- Agents, MCP servers and AI usage surfaced from the traffic itself
- Including the shadow ones no registry knows about
Where we fit — complement, don't replace
Your AI gateway & DLP / CASB
Route and log what AI does.
- Routing, fallbacks, cost control, caching, quotas
- Access, observability, request logging
- DLP on files, web and email
- Keep all of it — it's not our job
BlueAspen
Inspect and control what the data does.
- Validated DLP on the result leg, redacted in place
- Fail-closed on ingress — a call we can't judge never reaches your data
- Inside MCP, A2A and model traffic — where DLP/CASB can't reach
- Discovery of undeclared agents, MCP servers and AI use
Where the leak is
An agent calls an EHR tool and the response comes back carrying a patient's SSN and record number. An employee pastes a draft deal memo into a public chatbot. In each case BlueAspen inspects the data in flight — redacting the sensitive fields in place on the result leg, or refusing a call it can't judge before it reaches your data — and records every decision with a reason an auditor can read. 12 analyzers run on local ONNX models, so by default no customer data goes to a third-party LLM.
Built for regulated enterprises
Healthcare
PHI on every AI path
Clinicians using AI, and agents reaching patient data through MCP tools. Detect and redact PHI — SSN, MRN and more — before it leaves.
Fintech & insurance
PCI, account & model data
Card numbers, routing and account data validated with real check digits (Luhn, ABA, IBAN) — a reportable breach stopped, not just logged.
Two buyers, one engine
Land and expand
Platform & AppSec own the agent side; security & DLP own the employee side. Same product, same policy engine — expand from either.
Detects regulated data across both surfaces
PHI
PCI
PII
Secrets
Founded by enterprise-security leaders (ex-Proofpoint, ex-Workday, ex-Teradata), bringing data-loss prevention to AI.