Skip to content

Automatic translation — the Portuguese version is the reference. View in Portuguese

Updated on 2026-10-08

Platform overview

A managed data platform, built on hyperscale cloud infrastructure in São Paulo, with your data under your control — and the three layers Ingestia Data, ingestia.bi and Ingestia AI.

On this page (13)

ingestia.io is a managed data platform, built on hyperscale cloud infrastructure in São Paulo, with the customer's data under the customer's control. It brings together datalake, BI and AI for Brazilian companies:

  • Your data under your control: each workspace has its own datalake, isolated from the others. In the own cloud model (BYOC — included from Scale, add-on on Business), the datalake runs in your company's cloud project.
  • Availability by contract: SLA of 99.5% per month from the Growth plan, with service credit (Service level) and public measurement at /status.
  • Support: Aiô answers 24 hours, 7 days, inside the system (⌘/); human support works 9 a.m. to 6 p.m., Monday to Friday.

The platform is organized into three layers that you go through in order: data comes in through Ingestia Data, becomes analysis in ingestia.bi, and gains explanation and automation in Ingestia AI.

What it is#

A single path, from the raw file to the Monday-morning decision:

code
   source (file/ERP/Ads) people
          │ ▲
          ▼ │
   ┌──────────────┐ ┌──────────────┐ ┌──────────────┐
   │ Ingestia │──▶│ ingestia.bi │──▶│ Ingestia AI │
   │ Data │ │ (analysis) │ │ (explanation)│
   │ (data) │ │ │ │ │
   └──────────────┘ └──────────────┘ └──────────────┘
   landing + semantic model text-to-SQL, Q&A,
   Bronze/Silver/Gold visuals, dashboards alerts, narrative
   in the datalake and distribution (deterministic + LLM)

The three layers#

1. Ingestia Data — the Data layer

Brings the customer's data in and gets it ready for analysis:

  • Connectors and file upload feed the landing zone (the platform's file area).
  • The medallion pipeline materializes three tables per source in the datalake: Bronze (raw), Silver (clean and standardized), and Gold (datamart ready for BI).
  • Includes profiling with a masked sample, schema change detection (drift), scheduling, retry/DLQ, and a cost guard.

Documentation: Data (Ingestia Data).

2. ingestia.bi — the analysis layer

Turns Gold tables into analysis anyone can consume:

  • A central semantic model (relationships, measures, and formulas in Portuguese), reused by every dashboard.
  • A canvas editor with a catalog of 46 data visuals + 4 layout elements, filters, drill, themes, and versioning (N−1 decoder: an old dashboard never breaks).
  • Distribution: secret link, embed, open web, email subscription, and export (XLSX/CSV/PDF).

Documentation: BI (ingestia.bi).

3. Ingestia AI — the explanation and automation layer

Explains the numbers and automates monitoring. It has two very different families:

  • Deterministic (pure calculations running on our server): anomalies, forecasts, and alerts. They never send data to an LLM, zero token cost, bit-for-bit reproducible results.
  • With LLM (Anthropic): text-to-SQL, Q&A with citations, insights, and report narrative. They go through egress control, a capped budget, and decline when there's no basis — the AI warns you, it doesn't make things up.

Documentation: AI (Ingestia AI).

How the layers connect#

  • Data → BI: dashboards read from Gold tables (or from the semantic model grounded on them). Nothing in BI points to raw data.
  • BI → AI: the AI is grounded in the semantic model and the dashboards — that's where the citations and the provenance of the answers come from.
  • Everything scoped per workspace: each customer has its own isolated dataset; no layer leaks data between workspaces.

What it's for#

  • Small business owner: opens a dashboard and knows at a glance whether the month will hit target.
  • Analyst: connects sources, models, and builds dashboards without writing infrastructure.
  • Team: consumes dashboards (viewer), receives reports by email, and asks the data questions in plain language.

When to use it (and when not to)#

  • Use it when you want to move from scattered spreadsheets to a datalake with BI and AI, with the data in Brazil, paying for usage.
  • Pick the plan by the commitment you need: Starter is best effort; from Growth on there is a 99.5% monthly SLA. For stricter requirements (percentage, response times or own cloud), Enterprise is negotiated by contract.

Plan and permissions#

  • All three layers exist on every plan (Starter → Enterprise); what changes are the limits (connections, creators, viewers, refresh modes) — see Plans.
  • The owner manages structure, plan, and billing; roles and fine-grained permissions in Roles (RBAC).

Where your data lives#

  • Analytical data: dedicated datalake in the São Paulo region, one isolated dataset per customer (<prefixo>_bronze / _silver / _gold).
  • Files (landing): the platform's file area, in the same region.
  • Account / control plane: managed Postgres, also in Brazil.
  • Own cloud (BYOC): from Business (add-on) up, the datalake lives in your company's cloud project, in your account.
  • Security certifications (such as ISO 27001 and SOC 2) belong to the cloud provider that hosts the infrastructure.

Limits and costs#

  • Availability: 99.5% per month from Growth, tracked at /status. Cost values on screen are estimates; the amount charged is the one on the invoice.
  • Cloud usage (processing + AI) is paid in credits, with a per-query cost guard. See Usage and quotas.

Security#

  • Isolation per workspace (dedicated dataset, queries scoped by prefix).
  • MFA required on every account.
  • RLS/ABAC by member attribute; PII masked in samples and public snapshots.

Related#


Last reviewed: 2026-10-08.

Related links