43,000+
Government requests
Google fulfilled in 2024
Official Transparency Report. Majority without notifying the customer. Includes data from Brazilian users.
Google Vertex AI offers Gemini 2.5 Pro with a São Paulo region. But Google LLC is an American company — which means CLOUD Act, global retention policy and lock-in with Google Workspace. For regulated Brazilian companies, it's worth looking at the national alternative.
Metrics from Brazilian companies that migrated to MDA in the past 12 months. Anonymized Enterprise customer data.
Official Transparency Report. Majority without notifying the customer. Includes data from Brazilian users.
Gemini's cheap, but Vertex AI Search, Cloud Storage, Logging, Pipelines and BigQuery analytics add up. BRL 26k → BRL 4.4k/month.
Keep Workspace, BigQuery, Drive — MDA has native MCP Connectors. Exiting Vertex doesn't force you out of Google.
For companies in regulated sectors (finance, legal, health, government), these four axes determine whether generative AI is viable or a legal risk.
MDA is a 100% Brazilian company. CLOUD Act, FISA Section 702 and Executive Order 12.333 (US intel) don't apply. Google LLC is American — all of them apply.
Vertex hosts inference in São Paulo, but Cloud Logging is global by default. Cloud Identity is a global service. MDA keeps everything (inference + logs + identity) in a single BR region.
Gemini tokens seem cheap. But Vertex AI Search ($4/1k queries), Cloud Storage, Cloud Logging ($0.50/GB), Vertex AI Pipelines and BigQuery analytics multiply. ~BRL 26k/month for 50 users.
Native MCP Connectors for Drive, Gmail, Calendar, Sheets, Docs, BigQuery. You keep the entire Google ecosystem. Just swap the AI brain. Adapter for Vertex AI Pipelines included.
The 6 criteria that matter most to regulated Brazilian companies in 2026. No rhetoric — just verifiable data.
The MDA platform is model-agnostic. You can keep using Gemini 2.5 Pro for cases where long context (1M tokens) makes a difference — just routed through our gateway, with PII redaction before it leaves, unified audit log and cross-provider cost ceiling.
Real scenarios: long legal document analysis uses Gemini (1M token context). Customer service uses MDA 2.1 (sensitive data never leaves BR). BigQuery analytics uses Vertex directly. The same gateway logs everything, redacts PII before sending and applies the right quota.
CPF, RG, email, customer data redacted before any call to Google. Nothing sensitive crosses the border.
MDA + Vertex logs consolidated in one place — no need to open global Cloud Logging for internal audit.
Total quota per user/team/project regardless of provider. When it exceeds, automatic fallback.
Long document (>200k tokens)? Routes to Gemini. Rest goes to MDA. Automatic decision at the gateway.
MDA keeps Drive, Gmail, Calendar, BigQuery via MCP. Switching models doesn't force you out of Workspace.
MDA coexists with your Google Service Account. No data migration, no redoing permissions.
Compiled from 200+ real questions asked by CIOs, CTOs, DPOs and Heads of Data in MDA vs Vertex AI evaluation cycles.
Inference yes, logs/metadata no. Cloud Logging stores globally by default (you need to configure Regional Log Bucket), Vertex AI Pipelines has global aggregations and Cloud Identity is a global service. For LGPD compliance in regulated sectors, complete sovereignty requires a 100% Brazilian provider.
On multi-step reasoning and long context (1M tokens vs 256k), Gemini has an advantage. For PT-BR on typical enterprise use cases (customer service, document analysis, sales agents, BI), the difference is imperceptible — both solve at parity. For extreme context cases, MDA allows intelligent chunking that compensates for the difference.
Yes, and it's the most common use case. MDA has native MCP Connectors for Drive, Gmail, Calendar, Sheets, Docs. You keep Workspace and use MDA as the AI layer. No email or document migration.
No. MDA has a BigQuery connector that executes queries on-demand without replicating data. You decide which datasets are accessible to MDA via permissions (Service Account with restricted scope).
On average 70-83% for companies with 50+ users. Tokens seem cheap, but Vertex AI Search, Cloud Logging, BigQuery analytics add significant bills. Real case: company with 60 users on Vertex paid BRL 22k/month; after MDA Growth, pays BRL 5.4k — BRL 200k/year in savings.
Yes. MDA 2.1 LLM supports image analysis, document and PDF OCR, and image generation via partnership with Brazilian models. Vision and document understanding have parity with Gemini 2.5 Pro in internal tests.
Via Google Service Account with restricted permissions. MDA calls BigQuery, Cloud Storage and other GCP services on-demand, with automatic PII redaction before processing. Configuration done by the MDA team during implementation.
You can keep the CUD for other GCP services (compute, storage, BigQuery) — you just stop consuming Vertex AI. There's no penalty for reducing Vertex usage alone.
Teams using Google Workspace, BigQuery and GCP hesitate to migrate AI for fear of breaking the rest of the architecture. You don't have to. MDA has native MCP Connectors for the entire Google ecosystem — Drive, Gmail, Calendar, Sheets, Docs, BigQuery, Cloud Storage. You keep everything, just swap the AI brain for a sovereign alternative.
In a 5-business-day technical assessment, we map your current Vertex usage (including Pipelines, Search and Cloud Logging), design a phased exit plan without disruption, and deliver a realistic 4-week migration timeline.