Simat Labs
Book a demo

End-to-end visual intelligence systems

Simat Labs conducts research and builds highly optimized visual intelligence systems to solve real-world problems. At Simat, we understand the high cost of computation and focus our efforts at reducing it.

Adaptable

In the field

Visual intelligence at work

Real footage from systems we have built and deployed: safety, counting, crowd flow and more, each running on ordinary cameras.

  • Camera footage: Walkway safetyWalkway safety
  • Camera footage: Loading-bay countingLoading-bay counting
  • Camera footage: Crowd heatmapsCrowd heatmaps
  • Camera footage: Forecourt monitoringForecourt monitoring

How we work

End to end

From site inspection to deployment. We handle the entire pipeline.

  1. 01

    Site inspection

    We walk your site, check the cameras, lighting and network, and agree on what the system has to get right.

  2. 02

    POC delivery

    We build a working proof of concept on footage from your own cameras, so you see real results before you commit.

  3. 03

    Cost and sizing

    We work out the cameras, hardware and compute the full system needs, and give you a clear cost.

  4. 04

    System development

    We train the models on your real conditions and build the full system around them.

  5. 05

    System optimization

    We tune every model to run fast on the hardware it will live on.

  6. 06

    Deployment

    We install the system, connect it to your tools, then keep monitoring and improving it.

Deployment

On edge? On cloud?

We offer different deployment options, tailored to your needs and infrastructure.

  • 01

    On-premise

    Runs on a server inside your own building. Video never leaves your network.

  • 02

    Edge

    A small computer beside the camera. Results in milliseconds, and only events leave the site.

  • 03

    Cloud

    Cloud GPUs take on heavy jobs and scale across many sites.

  • 04

    Hybrid

    Mix and match: edge where speed matters, on-premise where video must stay inside, cloud for the heavy jobs.

  • 05

    BYOC

    Bring your own cloud. We deploy inside your own cloud account, so data, access and billing stay under your control.

Workflows

Build workflows. Get alerted.

IFA is our modular video AI platform. Each computer vision solution plugs into IFA and sends it events. You build workflows that turn those events into actions.

Walkway violationEvent from Walkway safety
Alert safety team
Daily safety report
Email Ahmad
Email managers
Pause Line 3
Send WhatsApp

Analytics

Generate valuable insight.

Use the events your computer vision solutions send to build anything from simple counts to complex analytics.

People counted

86,928

▲ 11.5%vs previous 30 days
Oct 17Previous periodNov 15

Research/ Distillation

Same answer.About 30× sooner.

We distilled Qwen3-VL-8B into a small model fine-tuned for a computer vision task. It matches the 8B model's accuracy and returns each answer in about 16 ms.

Latency per video window, batch 1, one A100

Qwen3-VL-8Breads every frame
513 ms
Simat Labs distilled modelsame accuracy
16 ms
faster than Qwen3-VL-8B on computer vision tasks, at the same accuracy
~30×
throughput on Qwen3-VL-32B from our 4-bit quantization
+59%