A Nature-first, scientific and technology companyContact us
Taramicro
An interface to Intelligent Systems

Model Studio

Engineer cyber-physical systems, AI and APIs.

Right TechnologiesAssets and PatternsSystems Engineering
How we buildSee what we ship
Model Studio THE PROCESS PLANBlueprintLAYER 01FoundationLAYER 02Domain PlatformLAYER 03ApplicationLAYER 04Runtime
Build Feedback

A question that motivates us

How do you build intelligent systems for the physical world?

Systems that sense, decide and act in the real world are hard to get right. Three things influence the outcome.

Resources

Right Technologies.

Every system design requires a careful selection of resources. Model Studio enables selection of software, models, domain specific languages, ontologies, modeling and simulations.

Repeatability

Assets and Patterns.

Most systems are built once, for one site, by the people who understood that site. Very little carries over. The next project starts again from the beginning, and the cost never comes down.

Process-driven

Systems Engineering.

Most AI projects tend to run open-ended, because the real scope only becomes clear with the data and models. When the system specification is well defined, the timelines become more manageable.

Tell us about your next AI application.

Let's assess it with Model Studio and plan the next steps.

Contact us

Model Studio - an interface to Intelligent systems

Engineer cyber-physical, AI systems and APIs confidently.

Most tools pick one approach — rules, or machine learning, or a language model. We use them together. Model-driven applications running at the edge with explainable engineering principles.

Concepts and logical view

A map with intelligence increasing upward and determinism and explainability increasing to the left. Conventional digital transformation — IoT, DAQ, SCADA and BMS — sits low and highly deterministic. Classical methods sit above it: statistical process control, machine learning, ontologies, graph analysis, expert systems, numerical solvers, modelling and simulation. Transformer-based AI sits high on the right, less deterministic and more costly. Composing the two produces AI applications in the upper left: better explainability, determinism, cost and utility.
Read it as a direction of travel. Conventional automation is dependable, but it is not intelligent. Transformer-based AI is intelligent, but it is harder to explain and costs more to run. What a plant or a building actually needs sits in the upper left — intelligent, still explainable, and affordable to keep running. Getting there means using both together.

The process, end to end

  1. PlanBlueprint

    The solution set out layer by layer, with its tests and acceptance criteria.

  2. Layer 01Foundation

    The shared concepts every domain platform is built on.

  3. Layer 02Domain Platform

    Models, semantics and I/O for one domain — e.g., HVAC, Automation.

  4. Layer 03Applications

    Built on the platform. Enables rapid application development.

  5. Layer 04Runtime

    Where the application runs, and where people actually use it.

Runtimes - Change the model to switch the applications

Browser · WASM
The full stack runs in the browser.
Server · on-premise
Runs entirely inside your own network.
Android · Linux tablets
Operator and field interfaces.
Single-board computers · edge
Packaged hardware, deployed next to the asset.

Models, Software, APIs -> Hardware -> SoC

What makes it different - Establishing the process

Layered architectureFoundation → Domain Platform → Application → Runtime Commonly Everything bundled into a single layer. Model Studio Foundation, domain platform, application build and runtime are kept separate. That separation is where the flexibility comes from.
Reasoning enginesClassical, graph and transformer-based Commonly One approach — rules, or machine learning, or a language model. Model Studio Methods you can inspect and AI agents work together inside one architecture.
On-premise tooling Commonly Cloud deployment assumed. Model Studio Runs fully on your own hardware, including in the browser through WASM, on a mix of open-source and custom tooling.
Domain languages and semantics Commonly Generic, one-size-fits-all data models. Model Studio Each platform carries its own domain language and semantics, tuned to the domain it serves.
A process, not just tooling Commonly Highly opinionated tools. Model Studio Environment, process and packaging are delivered together, aimed at where the system will actually run.
Extensibility Commonly Unrestricted modification. Model Studio Extensions are added within defined bounds, so a change cannot quietly break the reasoning underneath.

Applications and case studies

Build applications tuned to your domain.

Applications (models) provide truly intelligent cyber-physical systems unlike traditional IoT applications. The process and platform underneath make the difference.

Many applications with unified process

Application capabilities

  • Telemetry
  • Asset monitoring
  • Alerts
  • Audits
  • Quality control
  • Diagnostics
  • Location-enabled services
  • Training

Industrial

Automation, HVAC, HMI

  • Decision support above existing SCADA and historians
  • Supervisory applications for PLCs
  • Fault detection and energy optimisation
  • Operator screens with model-driven advice

Instrumentation

Measurement and control

  • Supervisory measurement and control
  • Smart metering
  • Process analysers and transmitters
  • Diagnostics at the signal, on the edge

Agri-tech

Growing and energy

  • Agri-input channels
  • Water management
  • Garden management
  • Heat pumps and bio-energy

Case study — buildings

Product · delivered

HVAC Decision Support System

View HDS

HDS turns the building data from the BMS, the historian, or the sensors directly into intelligence insights that an operator can act on.

  • What is driftingWhere the plant has moved away from how it was commissioned, and by how much.
  • What is about to failEquipment losing performance over time, early enough to schedule the work rather than react to it.
  • Where energy is goingWhich patterns of use account for the consumption, and which of them affects the human comfort.

An expert system provides the recommendations; the operator stays in control of the plant.

Edge intelligence stack · composed in one application

  1. 01Statistical process control
  2. 02Machine learning
  3. 03Graph analysis
  4. 04Expert systems

Runs fully in the browser via WASM, and on a single-board computer at the plant room. Live demo: hds.taramicro.com

Case study — industrial automation

Reference Architecture

Alignment with ISA-95

Taramicro does not replace the automation already in a plant. Instrumentation stays as installed, and the control layer keeps controlling. What gets added is a layer of interpretation — supervisory applications above the PLCs, add-on modules alongside the SCADA and historian, and decision support at MES and ERP.

The ISA-95 automation pyramid, split down the middle. The right half is existing instrumentation and automation. On the left, Taramicro interfaces sit at Levels 1, 2 and 4 — supervisory applications for PLCs, add-on modules to the existing SCADA and historian, and decision support at ERP — with Taramicro intelligent controllers at Level 3, MES. Level 0, sensors and signals, stays as installed.

Both products above sit inside this picture. The temperature analyzer works at Levels 1 and 2; HDS reads from Level 2 and advises at Levels 3 and 4.

Case study — Agri-tech, Food and Beverages

Product · SKU TM-THERM-CGA-SMG01

Model-based Temperature Analyzer and Transmitter

Releasing soon

A programmable intelligent analyzer that sits between the signal and the control system. It takes the temperature signal alongside what the PLC (or any control element) already uses, runs its detectors (functions) at the edge, and passes results on to SCADA, to an HMI, or to a user smart phone.

The PLC continues to control. What the analyzer adds is interpretation — the difference between a reading that is out of range and a process that is quietly drifting.

Detectors (functions) run on the edge device itself to reduce latency in the analysis.

Benefits

  • Digital assetsPreparing datasets for process, equipment and quality analysis.
  • Energy and consumablesDetect patterns in the process data to optimize the resource usage.
  • An intelligent layerAugment the existing automation infrastructure with edge intelligence.

Packaged on a single-board computer for the panel, with mobile and HMI interfaces. Sits at ISA-95 Levels 1 and 2, alongside the PLC and the existing SCADA.

Tell us about your next AI application.

Let's assess it with Model Studio and plan the next steps.

Contact us