PII and IP never leak
Built-in Guardrails ensure CPF, CNPJ, health data, and intellectual property never appear in prompts or responses — output auto-redacted.
CPFCNPJCardEmail+30 entities
Deploying generative AI shouldn't mean giving up control. At MyDatAgent, security isn't a feature request — it's our core architecture. We've built the most robust ecosystem on the market so your company can harness AI's power with the zero-risk security your CISO demands.
Strong passwords aren't enough: every team member and every agent accesses only what they strictly need. Onboarding and offboarding in seconds via your corporate IdP.
Integrate with your identity provider (Okta, Azure AD, Google Workspace). Employee leaves the company, access to AI and data drops instantly. Native SAML 2.0, OIDC, and SCIM.
End Shadow AI. Every developer, analyst, or agent gets a unique, traceable, revocable key. The master key stays protected in the MDA backend.
Define who can create agents, query sensitive data, or only view dashboards. The intern doesn't have the same view as the CFO. Customizable and auditable roles.
If an agent is processing critical data, you can't be in the dark. Monitor in real-time every call, every token, every MCP tool triggered — and get alerted before the incident.
Every API call, every token processed, every MCP tool triggered. You know exactly what the AI is doing — and what it tried to do and was blocked.
Agent tried to access a table out of scope? One user's costs spiked overnight? Instant alerts via Slack, PagerDuty, or email — before it becomes an incident.
Full trace of who accessed what, when, and what the response was. Cryptographically signed logs exportable to your SIEM (Splunk, Datadog, Elastic). Ready for audit.
Compliance can't depend on model goodwill. MyDatAgent policies run via LiteLLM in real-time: if the rule is "SDR agent doesn't discuss legal", it doesn't. Period.
Built-in Guardrails ensure CPF, CNPJ, health data, and intellectual property never appear in prompts or responses — output auto-redacted.
The SDR Agent doesn't give legal advice. The BI Agent doesn't discuss competitors. Real-time blocking via LiteLLM, with versioned and auditable policy.
Your data is never used to train third-party models. Infrastructure runs in isolated VPCs 10.20.0.0/16 in Brazilian datacenters. 100% LGPD compliant.
The highest global compliance standards — so companies in highly regulated sectors (Finance, Healthcare, Legal) operate with peace of mind.
| Certification | Status | Report access |
|---|---|---|
|
SOC 2 · Type IAICPA · Point-in-time control design audit
|
Certified | Available under NDA · Enterprise plan |
|
SOC 2 · Type IIAICPA · Continuous control audit over 12-month period
|
Certified | Available under NDA · Enterprise plan |
|
ISO/IEC 27001Information security management system · international standard
|
Certified | Available under NDA · Enterprise plan |
Compliance isn't optional. With MyDatAgent, your AI infrastructure is born within global auditor and regulator requirements — and the evidence package arrives ready for your CISO and Legal team.
How do we guarantee end-to-end security? Through our armored orchestration proxy. No stage is optional.
Unique API key issued via SSO. Identity validated against corporate IdP. Session expires automatically per company policy.
The proxy intercepts the request. Applies per-user quotas, rate limits (TPM/RPM), and checks access policies via RBAC. If the rule denies it, blocked before filtering.
Native Content Filter + Presidio + Lakera + Aporia run in parallel. Removes PII, blocks unauthorized topics, and prevents prompt injection before the model sees input.
SLM or LLM runs in your dedicated VPC (10.20.0.0/16) in a Brazilian datacenter. Zero leakage: data encrypted at rest and in transit, tenant isolation.
Who · when · which model · which prompt · which response · which MCP tool. Signed logs, hash-chained, and exportable to SIEM. Ready for audit and investigations.
The biggest barrier to AI adoption in banks, hospitals, and law firms isn't technology — it's governance. MDA Consulting bridges engineering and your CISO.
We map your regulatory requirements (LGPD, Bacen, ANS, OAB), configure policies, implement anomaly alerts, and deliver all necessary SOC 2 / ISO documentation and reports so your Legal and Security teams can green-light the project.