Explore data, detect anomalies, view operating regimes and keep monitoring as your data streams. It lives in your workflow and your everyday AI-agent.
Works with Cursor, Claude, and any MCP client. Built on published research. Hosted in the EU.
There is structure hidden in your data, and Barivia brings it to light. Explore it in plain language, see the distinct states your process runs in, and catch the readings that start to drift.
It works on structured data such as sensor streams, time series, and spreadsheets. You get visual maps and clear summaries, not a black box, so you can see what normal looks like and the moment something moves away from it.
Connect it once and it keeps watching as new data streams in. The first picture is where you start, not a one-off report you have to commission again.
The map is not the territory. Though a better map changes everything you can do with it.
Start with exploration, add continuous monitoring, or verify CFD meshes. They share the same engine and the same way of working through your AI agent.
Explore your data, see the distinct states your process runs in, and catch the anomalies and regimes as they appear.
Keep watching live data once it is set up, and get alerts the moment readings drift from normal.
Verify mesh convergence at field level across a refinement study, alongside classical Richardson and GCI.
The patterns are there. Standard tools may not see them. Here are the domains where we see the most potential.
Generation, distribution, and consumption data. Load patterns, anomaly detection.
High-frequency sensor streams. Operating modes, process monitoring, failure precursors.
Market regimes, cross-asset dynamics, temporal structure in price data.
Customer segmentation, operational patterns, behavioural analysis.
Weather patterns, seasonal cycles, spatial clustering in monitoring data.
Spectral signatures, classification, feature extraction from complex data.
CFD mesh-convergence verification and grid-convergence analysis for refinement studies.
Barivia Analytics Engine
Bring structured, tabular data such as a CSV or a spreadsheet. Connect through your AI agent, and the engine tells you the shape it needs, with no heavy preprocessing.
Interactive maps, cluster patterns, and plain-language summaries, ready within minutes.
Export results into your workflow, run deeper follow-ups, or share findings with your team.
Barivia lives in your everyday AI agent. Connect it in one command, explore your data in plain language, then let it keep watching through our REST API once you move to continuous monitoring. You stay in the loop while exploring, and it runs on its own once it is set up.
Connect in one command, works in Cursor, Claude, and any MCP client.
The CFD Mesh-Convergence Toolkit connects the same way, with @barivia/barmesh-mcp.
AI agents can discover the tools and connect automatically.
Add an MCP server that runs via npx against api.barivia.se with your API key. We send the exact package name and configuration when you start a pilot.
Upload your data and tell the AI what you're looking for. No code required.
The Analytics Engine trains, analyzes, and visualizes. Maps, regimes, and metrics emerge. Or run continuously: agents project new data and surface alerts automatically.
Works with
CFD Mesh-Convergence Toolkit
Compare the meshes of a refinement study by the distribution of their cell data, not just a single scalar, for a field-level read on whether your mesh has converged. Complements classical Richardson/GCI.
See where meshes still differ across the whole field, not only at one probe point.
Richardson extrapolation and the Grid Convergence Index on your scalar quantities of interest.
Upload a per-cell CSV and run it through the MCP; figures and a convergence summary come back.
Available via the Barivia MCP for teams running CFD refinement studies.
Continuous Monitoring Framework
When connected to live data streams, the platform monitors processes continuously. Trained on historical data, it projects new readings in real time and surfaces alerts the moment something shifts.
An alert surfaces the moment something deviates from normal behaviour.
Watch patterns form and shift as new data arrives, not after the fact.
Automated updates so your understanding stays current, not stale.
Production lines, energy grids, sensor fleets. Wherever data flows, continuous oversight follows.
Start free, then add capacity and capabilities as you need them. The same engine powers every plan.
Evaluate on your real data
Custom capacity and control
Pick a guided setup, then adjust in your dashboard if you need something different.
Self-Interacting Map and imputation when data is sparse — a common place to start.
Start with map + imputationAdd only the capabilities you need — open the dashboard and tailor the plan.
CFD mesh-convergence
Field-level mesh-convergence and Richardson/GCI for refinement studies.
Add in dashboardAnalytics engine
One engine for anomaly detection, regime analysis, and pattern discovery. Built on published research, not off-the-shelf tools.
Performance
An optimized Julia core with GPU-ready paths. Large datasets finish in minutes, not hours.
Infrastructure
EU-hosted compute, Cloudflare edge, and PostgreSQL persistence. Production-grade from day one.
Data security
Encrypted at rest, TLS in transit, and per-tenant isolation. Your data stays in your scope.
Monthly or annual billing available. Prices exclude VAT where applicable.
Clarity from complexity. A platform you run, not a report you commission each time.
References on request.
EU hosting, Cloudflare, PostgreSQL, Redis. Built for reliability and scale.
Encrypted at rest, TLS in transit, and per-tenant isolation. Data is processed in the EU and is never used to train foundational AI models.
An engine that finds the structure in your data and turns it into visual maps and plain-language summaries. It surfaces anomalies, the operating regimes your process runs in, and how things shift over time.
No. You work through your AI agent in plain language. The engine picks the methods, tunes them, and hands back clear maps and summaries, so you stay in control without needing the internals.
Structured, tabular data such as CSV files, spreadsheets, time series, and sensor streams. No preprocessing or reformatting needed before you start.
Yes. Connect the Barivia MCP server to Cursor, Claude, or any MCP client with one command. For continuous monitoring, the same platform is available through a REST API.
Your datasets and results are encrypted, stored in the EU, and isolated per organisation. They are never used to train foundational AI models.
Start on the free tier to evaluate on your own data. Paid plans add GPU compute and capabilities, billed monthly or annually. Contact us for early access while we onboard customers gradually.
