Today, the robot can autonomously sort objects such as bolts and nuts, achieving a successful cycle rate of over 75% on our lab test bench. The policy was trained on human demonstrations collected through teleoperation.
We have also developed our own Crimson Telepresence software for remote robot control, live camera streaming directly in a browser, and trajectory recording for training. Live demonstrations of the current capabilities are available on request — you can also propose your own operation as a pilot use case.

Conventional industrial robots are highly effective at stable, high-speed operations, but they usually require dedicated tooling, programming, and workstation modifications whenever the task changes.
Crimson is designed for a different approach: two anthropomorphic arms work with existing tools and equipment, while new operations can be learned from human demonstrations rather than programmed manually.
The product line includes three form factors for different types of tasks: a system on a movable stand, a stationary robotic cell, and manipulators mounted on a mobile platform.

Minimizing changes to the existing infrastructure is one of the core goals of the Crimson platform. The robot is designed to work with existing equipment and tools.
The exact requirements — placement, power supply, network connectivity, and safety measures — are determined during a site assessment. Based on the assessment, you receive a proposed layout, risk evaluation, and transparent cost estimate before making any commitments.

Our target is to retrain the robot for a new operation within one day using a learning-from-demonstration approach: an operator demonstrates the task through teleoperation, and the system learns from the recorded data.
The exact timeline depends on the complexity of the operation and is validated during the pilot. We define success criteria in advance and measure the time required to collect demonstrations, train the model, and achieve stable task execution.

No. After basic training, your existing employee will be able to start, stop, and monitor the robot without any programming skills.
During pilot projects, training the robot for new operations is carried out by Crimson specialists together with your representative who understands the production process. Teleoperation is used only to demonstrate new tasks and handle complex cases — not for continuous robot control.

This is exactly what Crimson is designed for: one platform that can be adapted to multiple tasks instead of deploying a separate robotic system for every operation.
Changing the operation does not require replacing the robot. The system is reconfigured by selecting the appropriate gripper or tool, demonstrating the new operation through teleoperation, training the model, and validating the result against measurable acceptance criteria.
The time and cost of reconfiguration are estimated for each operation and compared with alternative solutions before the work begins.

Our target unit price is 50k USD excluding VAT, with a target payback period of approximately 30 months when replacing one full-time position with annual labor costs of around RUB 20k USD.
These are target figures. The actual business case is calculated for your specific operation using real data on shift patterns, productivity, and total cost of ownership. The assumptions are then validated against measured pilot results before a scaling decision is made.

We are considering several flexible commercial models: direct purchase, leasing, and Robot-as-a-Service (RaaS) starting from a target price of 50 USD per day.
You can start with a low-risk pilot focused on a single operation. Commercial terms, success criteria, responsibilities of both parties, and exit conditions are agreed before launch. A purchase or scaling decision is made only after reviewing the final pilot results.

We have built and are operating three full-size demonstration units. Our teleoperation and data-recording software is already functional, and trained policies are currently being tested.
According to the roadmap, the next milestone is completion of the MVP Pilot in December 2026, followed by the first pilot deployment in a real production operation in January 2027.
We are currently building a pipeline of pilot sites, giving early customers an opportunity to become among the first adopters and influence the product’s development around their specific use cases.

Data ownership is defined contractually rather than unilaterally. Data categories — including video, telemetry, and production parameters — as well as ownership, storage location, and retention periods are specified in the pilot agreement and Data Processing Agreement (DPA) before deployment.
Customers may prohibit data transfer or the use of their data for training general-purpose models. Remote access to the robot is enabled only with the customer’s written consent and can be disabled at any time.