Sagi Ben Moshe and the cameras that learned to pay attention

A security camera is a remarkably patient employee. It can spend an entire night watching a door and still require someone else to explain what happened. Sagi Ben Moshe has built his latest company around that small absurdity: organizations collect video continuously, yet useful understanding often arrives only when somebody starts looking for it.

Ben Moshe is the founder and CEO of Lumana, which applies artificial intelligence to existing video feeds. Its proposition begins with an ordinary piece of installed equipment. Keep the camera, then give the footage a more useful destination. The ambition is to make an event recognizable and actionable while there is still time for a person to respond.

In May 2026, he put the problem plainly: “Video data is not properly monitored or analyzed.” The sentence contains very little theater. It also describes a substantial gap between owning a camera and knowing what is happening in front of it. Recording a scene is an achievement of engineering. Deciding which part deserves attention is another.

His route to that question runs through a computer-science education, a startup acquired by Intel, years inside the chipmaker, and a return to founding. Across those changes in employer and scale, the technical subject keeps reappearing. Depth, movement, surroundings, context: Ben Moshe has spent his career helping machines make something usable out of the visible world.

Before the camera could understand

At the Technion - Israel Institute of Technology, Ben Moshe earned a bachelor’s degree in computer science in 2007 and a master’s in 2010. These are the academic foundations of a career that would repeatedly mix algorithms with hardware. The eventual products had to operate outside a classroom, where environments arrive without an instructor to tidy them up.

His early Intel experience included part-time work on the design team for the first mobile Intel Core 2 processor. He left to build Invision, serving as its founder and chief executive from January 2009 to January 2012. The startup worked on 3D technology, giving the abstract problem of computer perception a commercial form.

Invision brought him into partnership with Ron Kimmel, the Technion computer-vision researcher. The company’s work used 3D reconstruction to recognize and interpret human movement. A gesture, after all, is only useful to a computer once movement has become a signal it can process. The business sat where mathematical description meets a person doing something in a room.

Intel acquired Invision in 2012, bringing Ben Moshe back into the company. He had moved from an established engineering organization to a startup and returned with a business of his own behind him. That sequence is a more revealing introduction than a list of senior titles: he had already worked on both sides of the acquisition table.

A machine needs its surroundings

At Intel, Ben Moshe’s responsibilities grew to include RealSense, hardware engineering, and the Emerging Growth and Incubation group. He also served as senior vice president of sensing technologies at Mobileye, with responsibilities for EyeC LiDAR and radar. The common concern was perception, though each role put it inside a different organization and product challenge.

RealSense developed hardware and software that allowed devices to perceive their surroundings in three dimensions. For a machine, an image and an understanding of distance are different kinds of information. A robot navigating a space needs relationships between objects, surfaces and its own position. A pretty picture offers limited consolation when the robot meets a wall.

The work had practical outlets. In 2017, Clearpath Robotics announced TurtleBot Euclid, combining RealSense depth-camera technology, an inertial measurement unit and a computer running Ubuntu and ROS. Ben Moshe welcomed the opportunity for people with different levels of robotics experience to build with it. Perception was becoming something developers could use, rather than admire from a distance.

His RealSense leadership also involved the less photogenic work of turning technology into a product. Research and development, marketing, finance and sales came within his remit. So did product definition, budgets and long-term planning. A sensor can be demonstrated in seconds; getting it into production calls for a much longer conversation with many more people.

The first RealSense-enabled camera, introduced in 2015, earned him an Intel Achievement Award. He has also been granted multiple patents. Those details describe tangible results of the work without resolving the larger challenge that would follow him: the distance between producing visual information and helping someone make a decision with it.

Leaving, with a team in the sentence

In 2018, Ben Moshe completed the Stanford Executive Program. By the end of his Intel tenure, his roles included corporate vice president and chief incubation officer. Incubation meant dealing with businesses before their eventual shape was settled, a useful vantage point for someone who would soon return to the uncertainty of starting a company.

When he announced his departure in 2021, he looked back at work his team had done together. He listed three Intel Achievement Awards, the formation of the incubation group, and ventures including drone light shows and geospatial work. It was a departure message with a varied inventory: computer vision had led well beyond one camera product.

“I’ll be missing my team,” he wrote. The remark is modest, but it gives the career a human scale. Executive biographies generally make organizations look like a succession of appointments. Here was the person leaving one, pausing over the colleagues who had made the work possible. The next venture would require assembling another group.

He became Lumana’s founder and CEO in August 2021. A subsequent recruiting post, using the Lumix AI name, invited people interested in visual intelligence to look at open roles. He also encouraged those who could not find a suitable position to reach out. Before the public launch and the funding headlines, there was the familiar founder’s task of finding people to build with.

“I’ll be missing my team.”

Sagi Ben Moshe, on leaving Intel in 2021

The lens can stay

Lumana emerged from stealth in April 2024 with $24 million in seed funding from Norwest and S Capital. Its hybrid-cloud approach combined video management with AI analysis. Ben Moshe’s new starting point was the camera already in place, with all the installation decisions, cables and procurement work that had brought it there.

The continuity with his earlier career is easy to see. RealSense helped machines perceive space. Lumana applies computer vision to footage organizations already produce. The emphasis has shifted toward interpretation and workflow: finding an event, sending an alert, supporting an investigation, and helping an operator decide what should happen next.

Kimmel is part of this chapter too, as Lumana’s chief scientific officer. Ofir Mulla serves as chief technology officer. The connection with Kimmel reaches back to Invision, making the new venture a continuation of an established technical relationship as well as a fresh company. Repeated collaboration gives the career a shape that funding announcements alone would miss.

Lumana’s capabilities include recognizing objects and behaviors, counting people, and identifying entry into designated areas. Video search and remote access bring another practical dimension. The point of collecting visual information is weakened if finding one relevant moment still requires an operator to inspect everything surrounding it. Search becomes part of the perception problem.

There is a quiet appeal in building around installed equipment. An organization’s camera network is already a collection of viewpoints on its daily life. Ben Moshe’s wager is that those viewpoints can support better decisions when software makes their contents easier to interpret. The lens remains where it is; the work expected of its output expands.

What 50,000 cameras does to the question

In July 2025, Lumana announced a $40 million Series A led by Wing Venture Capital, with Norwest and S Capital participating. Total announced funding reached $64 million. Wing founding partner Peter Wagner was to join the board. The financing supported further product development and the commercial work of bringing the platform to more organizations.

By December 2025, Lumana said it had surpassed 50,000 cameras. Its announced customer list included Meta, McDonald’s, New York University and the Minnesota Twins. Those names suggest varied settings for the same underlying problem. An institution, a restaurant operation and a baseball organization all generate footage, but the events their teams need to find will differ.

The number also changes the scale of the founder’s original question. One camera can be checked by one person. A large deployment demands a system that can sort relevant activity across locations and put it within reach of the people responsible. The work of attention becomes organizational as well as technical: who receives the result, and what can they do with it?

Lumana’s year-end account listed 65 new features and updates shipped during 2025, including developments in visual recognition and investigation tools. Software in this setting has to meet daily habits, rather than merely present a convincing demonstration. An alert earns its place when it helps the person using the system get through the next task.

An old question, a wider field of view

Ben Moshe’s responsibilities now extend beyond Lumana. He has served on Tower Semiconductor’s board since May 2024 and on its audit committee since October that year. That position adds a governance role to a career rooted in building and managing technology. It also keeps him connected to the semiconductor industry in which much of his earlier work took shape.

In May 2026, Lumana appeared among Qumra Capital’s eleven Tomorrow’s Growth Companies. Ben Moshe described a goal of bringing together multiple layers of information about the physical world. It is an expansive aspiration, but its practical starting point remains recognizable: footage that exists, an event inside it, and a person who could use an answer.

The thread through his career is the persistence of that problem. Invision interpreted movement. RealSense gave devices depth perception. Lumana asks existing cameras to contribute to monitoring, investigation and response. Different products, different organizational settings, and the same insistence that a visual signal should lead somewhere useful.

A camera can wait all night without getting bored. The person responsible for its footage has other things to do. Ben Moshe has built his current venture around respecting that difference. After years spent helping machines see, he is working on what their seeing should accomplish - and how much less footage a human should have to watch to get there.