— Convex Robótica

AI
for the
physical world.

AI, computer vision, sensing and robotics connected in a single architecture to observe, interpret and act on real-world operations.

Perception
Vision + sensors
Intelligence
AI + analysis
Action
Robotics + operation
CONVEX SYSTEM ARCHITECTURE / 01
Physical Intelligence Stack
CORE
Convex
intelligence layer
01 / DATA
Data
02 / PERCEPTION
Perception
03 / INTERPRETATION
Interpretation
04 / ACTION
Action
Data → Perception → Interpretation → Action
SYSTEM / ACTIVE

System

From signal to action.

A single architecture connects observation, context, interpretation and actuation — each stage feeds the next.

Conceptual scheme · schematic reading, not to scale
  1. 01

    Observe

    Sensors and imagery describe the environment and record the state of the operation.

    Satellite / drone / camera / sensor / weather
  2. 02

    Understand

    Computer vision, time series and context organize the reading of the signals.

    Computer vision / time series / context
  3. 03

    Decide

    AI, analysis and prioritization point to what deserves action first.

    AI / analysis / prioritization / report
  4. 04

    Act

    Operations, robotics and inspection execute in the physical world — and produce new records.

    Operations / robotics / inspection / intervention
ObserveUnderstandDecideAct

The records of each action return as data. The cycle restarts with more context than before.

/ Intelligence and computer vision

Nexa M1Nexa Vision

From imagery to diagnosis, turning data into clearer decisions.

01 / SOFTWARE
Nexa M1 + Nexa Vision
Territorial intelligence and computer vision applied to the operation.
02 / CONTROL
Synapse Control
Missions, telemetry and coordination of autonomous systems.
03 / PHYSICAL
Axis + Atmos
Robotics and sensing connecting software to the real environment.

Ecosystem

One architecture for the physical world.

Five layers connect sensing, perception, intelligence and execution — each product occupies the layer where it acts.

01

Data

Orbital, weather and field observation feeds the layers that follow.

  • Sentinel-2
  • Atmos
  • Sensors
02

Perception

Imagery and signals become a structured reading of the environment.

  • Nexa Vision
03

Intelligence

Analysis and interpretation turn reading into diagnosis and recommendation.

  • Nexa M1
04

Operations

Planning, mission control and follow-up of every action.

  • Synapse Control
  • Niros
05

Robotics

Physical execution: inspection, intervention and continuous operation.

  • Axis
  • Teros SPX-R

Conceptual layers · products listed by layer of operation

We source and connect technology for people and businesses.

Products

Perception and intelligence, in the same ecosystem.

Two products structure the reading of the physical world — and a family of systems connects data, control and execution.

Perception · computer vision01

Nexa Vision

Reads imagery and orbital data, extracts indices and identifies patterns and anomalies in the monitored area.

Explore perception
Intelligence · interpretation02

Nexa M1

Interprets signals with context, prioritizes what matters and delivers diagnosis and report.

Explore intelligence
Other systems in the ecosystem
  • 01NirosFollow areas, maps, weather, reports, alerts, Nexa Vision and Nexa M1 from your phone.
  • 02Synapse ControlOur ground control station for planning drone and autonomous robot missions.
  • 03AxisOur robot for process automation in agriculture and livestock.
  • 04AtmosAPI/platform for integrating weather stations from multiple manufacturers.
  • 05Teros SPX-RAerial scout for rapid mapping
  • 06VexAutonomous Interception & Containment Platform

Perception + Intelligence

What vision sees, intelligence explains.

Nexa Vision and Nexa M1 operate independently and complement each other when connected: one extracts the reading, the other gives it context and direction.

Conceptual scheme · schematic reading, not to scale
Nexa Vision · perception
  • 01Imagery and scenes
  • 02Spectral indices
  • 03Patterns and anomalies
  • 04Time series
Nexa M1 · interpretation
  • 01Operational context
  • 02Explanation of patterns
  • 03Zone prioritization
  • 04Technical report
  • 05Decision support

Conceptual ecosystem integration · each product also runs on its own

Technology in different contexts

Three applications to explore how data, intelligence and operations can work together. Requirements and integrations depend on each project.

Precision agriculture — ILLUSTRATIVE EXAMPLEBANDS → NDVI → MASK. scene = satellite.read("field.tif"); NIR, Red = scene.bands("NIR", "Red"); ndvi = (NIR - Red) / (NIR + Red); vegetation = valid & (ndvi > 0.3)CONVEX / VISION LAB01BANDS → NDVI → MASK1 scene = satellite.read("field.tif")2 NIR, Red = scene.bands("NIR", "Red")3 ndvi = (NIR - Red) / (NIR + Red)4 vegetation = valid & (ndvi > 0.3)NIR / REDNDVIVEGETATIONILLUSTRATIVE EXAMPLECONCEPT / NDVI

Precision agriculture

Understand differences across the field

Imagery and field information help observe crop variability and organize monitoring over time.

  • Connect observations with field context
  • Support inspection planning
  • Use maps and records to guide analysis
Explore application
Industry — ILLUSTRATIVE EXAMPLECAPTURE → INFERENCE → DETECTIONS. frame = camera.capture(); model = vision.load("inspection"); detections = model.infer(frame); overlay.draw(frame, detections)CONVEX / VISION LAB02CAPTURE → INFERENCE → DETECTIONS1 frame = camera.capture()2 model = vision.load("inspection")3 detections = model.infer(frame)4 overlay.draw(frame, detections)OBJECT / 01REVIEW / 02OBJECT / 03FRAME → MODEL → OVERLAYILLUSTRATIVE EXAMPLECONCEPT / CV

Industry

Connect information to the factory floor

Observing environments and processes can support evaluation and coordination between people and systems.

  • Organize operational signals and records
  • Provide context for process evaluation
  • Plan control and robotics applications
Explore application
Territorial monitoring — ILLUSTRATIVE EXAMPLEDATE A → DATE B → CHANGES. before = imagery.read("date_a"); after = imagery.read("date_b"); a, b = align(before, after); change = abs(b - a) > thresholdCONVEX / VISION LAB03DATE A → DATE B → CHANGES1 before = imagery.read("date_a")2 after = imagery.read("date_b")3 a, b = align(before, after)4 change = abs(b - a) > threshold2025-04-012025-07-01CHANGE MASKILLUSTRATIVE EXAMPLECONCEPT / DELTA

Territorial monitoring

Follow the territory over time

Imagery helps explore areas and compare records to focus on changes that deserve field analysis.

  • Consult spatial information
  • Compare imagery across periods
  • Guide monitoring of areas of interest
Explore application

Four commitments that run through every project.

Proprietary technology

We develop our own architecture and configure it for each operation.

Metrics before rhetoric

Every pilot ships with comparable ROI, OEE and telemetry.

Local engineering

A proprietary technology stack designed for continuous operation, technical support and long-term evolution.

Human work, extended

Automation frees up time, it doesn't replace people — we design with that commitment.

→ 01

Diagnosis

We assess the operation, the monitored area and the problem to solve, then define the right architecture.

→ 02

Configuration

Nexa M1, Nexa Vision, Atmos, Synapse Control and Axis are combined based on the data, sensors and automation level the operation requires.

→ 03

Deployment

We set up monitoring, analysis, sensing, mission control and automation in a single operational architecture.

→ 04

Operation

The platform continuously tracks the environment, turns data into decisions and connects digital intelligence to execution in the physical world.

Computer vision, AI and robotics, from data to action.

We build an integrated architecture of artificial intelligence, computer vision, sensing and robotics to turn real-world data into diagnosis, decisions and execution.

How Convex deployment works, from diagnosis to operation

Technical diagnosis

Scope, constraints and success criteria defined together with the client.

Technical configuration

The Convex architecture is defined for the operation, with scope, timeline and risks mapped out.

Field pilot

Assisted installation, live telemetry and team training.

Operation

Ongoing operation contract, support and iterative evolution.

Robotic hand developed by Convex Robótica manipulating parts in an industrial environment

Every technology we build starts with a simple question:

What human work does it free up? The answer defines material, sensor, motion, cost. The platform handles the rest.

Physical world

Software ends where the physical world begins.

Software, perception and control only come full circle when they become measurement and action on a real environment. That is why the architecture does not end on screen.

  • Software
  • AI
  • Vision
  • Sensors
  • Control
  • Machines
  • Physical environment

From isolated sensors to an ecosystem that learns together.

  • 01Discrete manufacturing
  • 02Logistics & intralogistics
  • 03Precision agriculture
  • 04Mining & inspection
  • 05Energy & utilities
  • 06Education & R&D

Technology that works with you.

Automation to extend, not replace. Our architecture takes on the repetitive and the risky so people and operations can focus on what really matters.

Technical illustration of a robotic arm in a conceptual study

Research and development

Technology built to leave the lab.

Perception, control, hardware and robotics advance in the same process of investigation — what we learn in the field returns to the next generation of the system.

  • 01Software
  • 02Perception
  • 03Control
  • 04Hardware
  • 05Robotics
Explore the research

From analysis to operation.

Explore the platform products or talk to the team about your project's context, needs and possibilities.