Services AI & models Projects Taxi Contact Start a project →
EN RU SR

TELETEH / Applied intelligence

Intelligence.
Developed
in-house.

We train models for tasks where understanding, speed and control matter. Then we put them to work in real products.

Explore the models ↘
TRAINED IN-HOUSE. BUILT FOR PURPOSE.AI / TELETEH

v-Model2

COMPUTER VISION

Your cinema.
Now with depth.

See familiar video in a new dimension. v-Model2 turns 2D into stereo during playback. Our proprietary model estimates 4K frame depth at 90 Hz with less than 50 ms latency. Press play and experience depth, without pre-conversion. Works on iOS and Android phones, as well as TV boxes.

  • On-the-fly depth estimation
  • Left and right view synthesis
  • Integrated into XPlayer2
XPlayer2 ↗
4Kvideo resolution
< 50 msprocessing latency
90 Hzdepth processing

GeoEnergo

ENERGY INTELLIGENCE

Spot the risk.
Act in time.

Test a decision before it changes the grid. GeoEnergo combines topology, telemetry and contextual factors to forecast line losses, overloads and demand growth. Before switching a feeder, see where and when the action could put infrastructure at risk.

  • Forecast line losses and demand growth
  • Assess overloads and infrastructure conditions
  • Warnings informed by location and timing
Scenariosbefore switching
Lossesline-level forecasts
Risksearly warnings

GisTT

SPATIAL INTELLIGENCE

Less reconciliation.
Faster decisions.

Turn plans, drawings and diverse sources into one working picture. GisTT automatically matches object properties and aligns layers. New urban planning and technical conditions (UTU), connection points and environmental impact calculations in seconds. Already working in eMapa.

  • Property matching and layer alignment
  • Connection points and new UTU calculations
  • Environmental impact assessment
eMapa ↗
Secondsper scenario
UTUconditions & connections
eMapadeployed in product

MidasTT

MARKET INTELLIGENCE

The market moves.
You are acting.

Shorten the path from market event to trading decision. MidasTT continuously analyses news and markets, produces signals and can act within your strategy. Its miniature architecture delivers millisecond inference for scalping and other latency-sensitive scenarios.

  • News and market context analysis
  • Fast signals for trading strategies
  • Decision automation within defined limits
msinference time
Compactmodel architecture
Event-drivenresponse to events

Vozzi Dispatch

MOBILITY INTELLIGENCE

The right vehicle.
For the right ride.

Our model for order assignment and taxi fleet analysis. It considers vehicle location, availability, pickup time and order context. It helps identify the causes of waiting, rejections and uneven utilization, supporting fleet decisions within the Vozzi white-label platform.

  • Context-aware vehicle matching
  • Pickup, rejection and utilization analysis
  • Integrated into the taxi platform
Vozzi ↗
Dispatchorder assignment
Fleetoperational analysis
White-labelunder your brand
Custom model training

Your challenge.
Your model.

When an existing solution falls short, we build our own. Or fine-tune an open model on your data. From dataset preparation to fast inference on your infrastructure.

Define the task

Agree on data, quality metrics, constraints and target hardware.

Train and evaluate

Prepare the dataset. Train from scratch or fine-tune. Evaluate against your scenarios.

Ship into production

Optimize speed and size. Deploy the API, monitoring and updates.

Your data. Inside your perimeter.

Proprietary and open models on your server or in your VPC. RAG over internal documents, integrations and team training.

Discuss private AI ↗
Discuss your project ↗