A modular approach to useful robotics.

Manort is building Milo, a modular semi-humanoid robot for physical work. The platform brings together a mobile base, a detachable upper body with robotic arms and hands, and AI software for perception and action.

Alongside the hardware, we are developing simulation and task-data tools to help customers prepare real workflows in a digital version of their own environment before deployment. Our initial focus is retail, inventory, and logistics, with household assistance as a longer-term ambition.

Retail and logistics teams face repetitive, physically demanding work and pressure on staffing. Automation can help, but it is often costly or difficult to adapt to a particular location and workflow.

We want to make adaptable robots practical in spaces designed for people. Over time, that could also mean helping older people carry objects, do housework, and manage everyday tasks so they can remain independent for longer.

Our goal is a platform that can support different kinds of physical work as its capabilities develop.

Milo separates mobility from manipulation. Its omnidirectional mobile base is being developed to work independently on transport, internal delivery, and material movement.

The detachable upper body adds arms and hands for tasks such as picking, packing, and shelf handling. The intended benefit is that customers can use the hardware a task requires, while sharing the same underlying platform.

The complete system is still in development. Integrating the base, upper body, docking, sensing, and software is the next stage of our work.

Our first target customers are retail chains and supermarkets, warehouses and logistics operators, and manufacturing or distribution teams.

Potential retail uses include moving products, supporting shelf work, and preparing inventory or fulfilment orders overnight. In warehouses and distribution facilities, we are exploring transport, picking, packing, and repetitive material handling.

Longer term, we aim to offer household assistance through a subscription model, giving people access to help with carrying, shopping, and housework. This is a future offering, with commercial workflows taking priority during early development.

Yes. Companies with their own compatible robot can explore a scoped project around Manort Simulator Engine and Manort Task Data. A project can begin with the customer’s site, a defined physical task, and suitable data they already control, or with a new task-capture project.

We then assess the environment, source rights, data quality, target robot, end effector, sensors, and control interface. The intended output is a Real-to-Sim Task Pack that connects structured demonstrations with a physics-based task environment and an agreed validation method.

This workflow is in development and is scoped individually; compatibility is established before a project begins. Explore Manort Task Data.

Manort sits at the intersection of professional service robotics and the emerging market for general-purpose humanoid robots. These categories cover different products and should be considered separately.

The International Federation of Robotics recorded more than 205,000 professional service robots sold in 2023, a 30% increase in its reporting sample. Nearly 113,000 were for transport and logistics. These figures come from a sample of 298 suppliers and are not an estimate of the entire industry. IFR, World Robotics 2024.

For humanoids, Goldman Sachs Research’s February 2024 outlook projected a global total addressable market of US$38 billion by 2035. Morgan Stanley’s May 2025 outlook projected an approximately US$5 trillion opportunity by 2050, encompassing a broader ecosystem of robots, supply chains, and maintenance and support. Goldman Sachs forecast · Morgan Stanley forecast.

These are forecasts with different scopes and time horizons, not current sales or estimates of revenue available to Manort. Adoption depends on progress in capability, cost, safety, and customer demand.

Relevant companies include Agility Robotics, Figure, Apptronik, and NEURA Robotics. Our intended differentiation is modular hardware and simpler deployment: separating mobility from manipulation, then helping customers build a physics-based version of their environment from images or 2D plans and prepare tasks through natural language.

Our long-term thesis is that more scalable manufacturing and easier use can make general-purpose robotics accessible to a much wider market.

We are combining a modular mechanical design with AI and simulation tailored to the places where the robot will work. Milo’s planned capabilities bring together autonomous navigation, LiDAR and depth perception, omnidirectional movement, docking, and manipulation.

Our simulation model converts images or 2D layouts into a physics-based environment. We are developing the workflow around it so users can describe tasks in natural language, test behaviour digitally, and prepare the robot for a location with less specialist programming.

As the platform develops, our aim is to improve it through operator feedback, validated actions, and experience from real workflows. That combination of hardware, deployment knowledge, and data is the technical advantage we are working to build.

Manort was founded by Saul Ortiz, Robotics Engineer, and Gulshan Mandal, AI Engineer.

Saul leads robotics, mechanics, and system integration. He began building robots in middle school and has participated in robotics competitions, including competitions connected with NASA.

Gulshan leads AI, simulation, and software. In 2026, he received Pforzheim University’s Research Excellence Award in the student category.

The founders bring complementary experience across physical robotics and artificial intelligence. Meet the team.

Manort has been fully bootstrapped. The founders have funded hardware, software, and early prototype development with their own private funds and retain full ownership to date.

We are open to strategic and financial investment from partners who share our ambition to make robotics more modular, accessible, and scalable. Our immediate focus is completing an integrated platform and preparing for customer validation.

We estimate that approximately 40% of the first working Milo prototype is complete. This is our internal measure of development progress as we build and test the full system.

The upper body and robotic hands have been built, with core mechanisms functioning. We have also developed our image/2D-to-3D conversion model for recreating environments in physics-based simulation.

Current work focuses on improving the hands’ reliability and manipulation capabilities while developing the autonomous mobile base. Physical hardware and simulation technology already exist; the complete system still needs integration and real-world validation before a customer pilot.

Our next major milestone is a complete, integrated Milo ready for a supervised customer pilot. That means improving the hands, completing the autonomous base, and bringing sensors, controls, power, docking, and software together into one reliable system.

We have spoken with a retail company in Germany and demonstrated how our simulation technology can turn a store layout into a 3D physics environment. We received positive initial feedback from this early customer conversation.

Our goal is to progress toward a retail proof of concept that can validate autonomous transport, docking, and manipulation in a real operating environment. Scope and timing will depend on technical readiness and agreement with a customer.

We are designing Milo around serviceable modules so components can be maintained, upgraded, or replaced without necessarily replacing the entire robot. The mobile base is also intended to perform transport tasks independently when manipulation is unnecessary.

Simulation can help us test workflows before repeated physical trials, with the aim of reducing unnecessary prototype iterations, travel, and material use. These are design objectives whose impact needs to be assessed as the system develops.

Our longer-term social aim is to support people with repetitive and physically demanding work, leaving more room for human judgement, interaction, and care. At home, we want practical assistance to support independence while keeping human relationships central.

We believe general-purpose robots can become accessible beyond large industrial companies. Our ambition is to make their hardware modular, their deployment simpler, and their interaction with people more intuitive.

We envision businesses configuring robotic capabilities around the work they need, and people teaching tasks through natural language without writing code.

We plan to begin with retail and logistics, expand into other commercial settings, and eventually explore household robotics through subscription access. The long-term goal is a scalable platform for physical AI that grows from focused deployments toward everyday use.