Product

Everything you need to turn data into decisions

Data sources, visualizations, dashboards, explainable machine learning models, alerts and generative AI, in a single platform hosted in the EU.

Data sources

Connect any data source

Bring your data from wherever it is. AIR-Core detects the columns and their types and keeps the sources up to date.

Files

CSV, Excel and JSON. Drag the file in and you're done.

URLs

Import data published on a website or returned by an API.

Databases

MySQL, PostgreSQL, SQL Server, Oracle, MongoDB and InfluxDB. Credentials are stored encrypted.

Open data

Connect CKAN open data portals.

Real time with MQTT

Receive data from sensors and devices as soon as it is produced.

Scheduled updates

Refresh sources automatically as often as you choose.

Derived and combined sources

Filter, transform and join several sources into a new one, without touching the originals.

Visualizations

Charts and maps in a few clicks

Choose columns, chart type and aggregation. AIR-Core does the rest, even with data arriving in real time.

BarsLinesScatterPiesRadarIndicatorsKPIHistogramsParetoHeat mapsTreemapVoronoi treemapChord diagramsWord cloudCountry mapsCity heat mapsTablesReal-time series

Dashboards

Dashboards to share what matters

Bring visualizations, indicators and text together in dashboards that update themselves.

Dashboard editor

Place and resize charts, indicators and text as you like.

Share with your team

Decide who can view or edit each workspace.

Public links

Publish a dashboard with a link for people without an account.

Embed

Embed dashboards and visualizations in your website or intranet.

AutoML models

Predictions that explain what they are based on

Choose what you want to predict or discover. AIR-Core prepares the data, compares algorithms, tunes their parameters and hands you the best model, ready to use.

Predicts which category each case belongs to.

AIR-Core tries several algorithms, tunes their parameters and keeps the one that generalizes best. You'll see its evaluation metrics and can ask for predictions one at a time or in batches.

ExampleWhich customers are most likely to leave this quarter?

Explainability

Models that explain their decisions

A model you don't trust is no use. AIR-Core shows why it predicts what it predicts.

  • Explanation of each prediction. Which variables raised or lowered the result, and by how much.
  • Global explanation. Which variables weigh most in the model as a whole.
  • Model card. Training data, metrics, limitations and out-of-scope uses.

Example · why high risk?

Months without buying
Open tickets
Tenure
Annual plan

Explanations are calculated with statistical methods (SHAP values) on our infrastructure; they do not use generative AI.

Alerts

Find out in time

Set conditions on your data and AIR-Core alerts you when they are met.

Conditions on your data

Set the condition on any source and AIR-Core watches it for you.

Email and webhooks

Get the alert by email or send it to another system with a webhook.

Explanations written by AI

Every alert comes with an explanation, in your language, of what happened.

Alerts in plain language

Describe the alert in your own words and the AI sets it up for you.

Generative AI

An assistant that works inside AIR-Core

Ask for what you need in plain language. The assistant looks at your data and creates sources, models and dashboards, always asking you to confirm before changing anything.

AIR-Core's AI, in the EU

Claude (Anthropic) through Amazon Bedrock with a European Union inference profile. AWS neither keeps the requests nor uses them for training, and we don't use your data to train models either.

Bring your own key

Use your Mistral AI (servers in the EU), OpenAI, Anthropic, Google Gemini or Groq account, or any provider compatible with the OpenAI API. Your key is stored encrypted and the provider bills you directly.

MCP server

Connect Claude, ChatGPT and other AI apps to AIR-Core through the MCP protocol. You decide the permissions and can revoke them whenever you like.

See the MCP server

AI creation wizards

Describe what you want and the AI fills in the wizard for sources, models or dashboards for you.

Alerts in plain language

Write the condition in your own words; the AI turns it into an alert and explains each notice to you.

AI credits

Paid plans include a monthly balance for AIR-Core's AI, with top-ups from €10.

MCP server

Use AIR-Core from Claude, ChatGPT or Cursor

MCP (Model Context Protocol) is an open standard for connecting AI assistants to applications. With the AIR-Core MCP server, your assistant works directly with your sources, models and dashboards, without copying and pasting data.

  • OAuth. You authorize the connection from your account, for the organization you choose.
  • Your permissions, no more. Read-only or read-and-write access, always within what you can do.
  • Audit. Every call is logged in the AI Activity.
  • No deletions. The MCP server offers no delete tools.
  • Hosted in the EU. It runs on the same AWS infrastructure in Frankfurt.

Available from the Starter plan. You can revoke access at any time.

Your AI assistantAIR-Core connectedSimulated demo
Which region sold most in October, and why?
AIR-Core · query_source_datasales_2026 · October · by region
AIR-Core · explain_predictionmodel “sales_region”
Answer (demo) The winner was North, with €412k (+18% on September). According to the model, the distributor channel and the autumn campaign weighed most. Norte412 k€Centro356 k€Este298 k€Sur241 k€Oeste187 k€
Every call is logged in the AI Activity.

What your assistant can do

Some of the tools the MCP server provides:

Data sources

List and search sources, see their structure, query data with filters and aggregations, create sources from a URL and derived sources.

list_sourcesget_sourcequery_source_datacreate_source_from_urlcreate_derived_source

Visualizations and dashboards

Create visualizations and dashboards, duplicate them and generate embed links.

create_visualizationcreate_dashboardclone_dashboardcreate_embed_link

Models

Train models, see their evaluation and card, predict in batches and explain the model or each prediction.

create_modelget_model_evaluationget_model_cardbatch_predictexplain_modelexplain_prediction

Alerts

Create alerts, look them up and review their history.

create_alertlist_alertsget_alert_history

Integrations

For technical teams, too

Everything you do in the interface can be automated.

  • REST API v1. Sources, models, predictions, dashboards and alerts, documented with OpenAPI.
  • Python SDK. Work with AIR-Core from your notebooks and scripts.
  • MCP server. For AI apps compatible with the Model Context Protocol.
  • Webhooks. Send alerts to your own systems.
# Prediction with the REST API v1
API=https://app.air-core.io/api/v1
curl -X POST "$API/workspace/$WS/models/$MODEL/predict" \ -H "x-auth-token: $TOKEN" \ -H "Content-Type: application/json" \ -d @customer.json

Organizations and security

Access control and data in Europe

Designed for organizations that work with sensitive data.

More about security

Organizations, roles and permissions

Group your team, assign roles and limit access per workspace.

Two-step verification

Protect every account with a second factor.

Hosting in the EU

AWS servers in Frankfurt (Germany).

Encrypted credentials

Your database passwords and API keys are encrypted at rest.

Deployment

In the cloud or on your servers

Get started in minutes with the cloud version. If your security or volume requirements call for it, we bring AIR-Core to your infrastructure.

AIR-Core in the cloud (SaaS)

Instant access from your browser, nothing to install. Choose the plan that suits you and change it whenever you like.

See plans

Private and on-premises deployment

Run AIR-Core on your own infrastructure or with your cloud provider, with a dedicated database, your domain and your brand, and a service level agreement.

Discover iCore

Start today with your own data

Create your free account and connect your first source in minutes. When you need more, move to a paid plan.