Dave Grannan Keeps Coming Back to the Human Problem

On the street, Dave Grannan had a reliable way of getting strangers to talk to him: take out his camera. The Light L16 had sixteen camera modules on its face. It looked as though someone had given a pocket camera an unreasonable number of responsibilities. People wanted to know what they were looking at. Grannan, the company’s co-founder and CEO, was happy to explain.

In a 2017 television demonstration, he put the device beside the equipment it was supposed to replace. A conventional camera, lenses, a backpack. The pitch was beautifully physical: here was all the stuff you might leave at home, and here was something you could bring. “What IS that?” was how he described the usual reaction. Before anyone judged the photographs, the object had already made an impression.

The L16 would become a difficult chapter in his career. It also makes a useful entrance to it. Grannan has repeatedly worked on the awkward space between what a machine can do and what a person can comfortably ask of it. Sometimes the obstacle is a tiny keyboard. Sometimes it is a bag full of lenses. Recently, it has been the telephone call nobody wants to wait through.

Before Siri, a smaller keyboard

Grannan began his career as a US Marine Corps officer. He later earned an MBA at Berkeley’s Haas School of Business and held leadership positions at Sprint PCS and Nokia. Those experiences put him close to mobile computing while its possibilities were expanding faster than its interfaces. Getting a device into a pocket was one achievement. Making it pleasant to use remained another.

By 1999, as president and CEO of Geoworks, he was working on software that connected mobile phones to the internet, email and desktop computers. Amazon bought a seven percent stake for $5 million. The familiar shopping company was making an investment in the less familiar business of helping information travel beyond a desktop. Grannan was already occupied with the journey from technical capability to ordinary use.

At Vlingo, where he became CEO in 2007, the question grew more direct: why should a phone owner have to learn a machine’s preferred vocabulary? Its speech interface let users speak naturally, rather than memorize a rigid menu of commands. Processing happened on remote servers, and the system could learn from corrections. Early work focused on feature phones, where entering words on a twelve-key keypad made the appeal particularly obvious.

Grannan’s explanation of the business was practical. People struggled to discover applications and struggled to use the ones they found. Vlingo’s BlackBerry software added spoken Facebook and Twitter updates and access to more applications. A phone could contain an impressive collection of services and still leave its owner hunting through menus. Speech promised a shorter route to what was already there.

The people who came back

Nuance announced its agreement to buy Vlingo in December 2011, after a fierce rivalry. The companies had fought; now their people would have to work together. Grannan’s public response acknowledged the conflict and looked toward joint work. By October 2012, he was a Nuance vice president discussing Dragon Mobile Assistant, an Android product that could send messages, make calls, set appointments and retrieve information through natural language.

The Vlingo chapter also left behind a set of relationships. Han Shu and Ghinwa Choueiter worked there, as did Joe Pirone and Chris Barnett. Years later, those names would reappear around Grannan at Marr Labs. Barnett also worked with him at Light. That continuity gives his career a texture missing from a list of company names: colleagues kept finding reasons to build together again.

His work extended beyond running companies. He chaired the board of Maluuba, the language and speech technology business acquired by Microsoft. Cognii lists him as an adviser and records board roles at Sensory and Sense. These are different applications of technology, from speech to home energy monitoring, but they put him repeatedly beside teams trying to turn complex information into useful decisions.

His professional profile also contains a small, pleasing detail: at Haas, he was editor-in-chief of Cal Business Weekly. Before the product demonstrations and fundraising conversations, there was a student newspaper. It is an apt footnote for someone whose job has so often involved explaining why an unfamiliar idea deserves a reader’s, customer’s or investor’s attention.

Sixteen eyes meet the real world

In 2013, Grannan founded Light with Rajiv Laroia. Their partnership brought an experienced technology executive together with an engineer whose work spanned communications and imaging. Light’s proposal was to combine the output of several small cameras, using computation to obtain capabilities normally associated with larger photographic equipment. The company had to make an optical idea, electronics and software behave as a single product.

Grannan described talking with photographers and filmmakers about their decisions before, during and after taking a picture. Asked what to expect next, he offered “Total world domination,” then returned to discussing applications. The joke suited the scale of the undertaking. Cameras had an established vocabulary and demanding customers; a new arrangement of lenses had to earn its place in both.

In July 2015, Light was moving from components to working prototypes. Funding had distinct jobs: the earlier round helped establish a prototype; the next would support the team and production. That is a less glamorous account of invention than a sudden flash of insight, and a more useful one. A photograph made in a development lab still has a long way to travel before a customer can reliably make one.

The L16 announcement that October drew an immediate response. Grannan said Light passed its first month’s preorder target in two days. Interest arrived before independent reviews. By the time the camera reached customers in 2017, however, the challenge had shifted from explaining why sixteen modules were interesting to proving that the entire experience justified carrying another device.

The camera’s price approached $2,000, and its image-processing workflow asked more of buyers than the original pocketable promise suggested. Hardware and software shortcomings mattered. So did the improving cameras already inside phones. The L16 offered a distinctive answer while the customer’s alternatives kept changing. For Grannan, consumer interest and a workable long-term business had proved to be separate hurdles.

A different use for depth

Light’s cameras could extract depth information from different views of the same scene. That opened a second question: what if the customer needed to measure the world rather than photograph it? During fundraising discussions, SoftBank’s Masayoshi Son encouraged an automotive application. Laroia spent months investigating whether the approach could work with widely separated cameras on moving vehicles.

Grannan subsequently acknowledged poor timing in smartphones: extra cameras and specialized processing added expense in a market increasingly concerned with price. Light moved its attention to vehicle perception. The change gave its underlying imaging work another potential customer, while demanding that the technology function amid movement, vibration and the practical constraints of a vehicle.

In 2022, Deere acquired Light-related depth-sensing and camera-perception technology. Its annual report lists LGT, LLC, identified as Light, among that year’s acquisitions. The story had traveled an unexpected distance from a photographer’s backpack. Agriculture had a use for machines that could perceive their surroundings. Grannan’s camera company had reached a buyer whose products spent their working lives in fields.

Back on the line

In 2023, Grannan co-founded Marr Labs; the company joined Y Combinator’s Winter 2024 batch. Han Shu became its co-founder and CTO, and Ghinwa Choueiter was another co-founder. The return to voice came with people who had helped build the earlier generation. The new product, Vox, targeted business telephone work, including appointment scheduling, sales and customer support.

Mortgage lending gave the team a specific task. Prospective borrowers needed a prompt response, while lenders had to manage changing demand and connect interested callers with qualified staff. Marr’s approach included asking initial questions and handing the conversation to a live loan officer. The handoff mattered as much as the voice. A natural-sounding exchange that ended in an abandoned call would still leave the job unfinished.

In an April 2025 discussion, Grannan described rapidly adding or removing AI calling capacity. He also emphasized routing callers to people when they asked, or when a conversation needed human attention. Those two ideas belong together. Automation offered flexible capacity; escalation offered an exit from it. The caller still needed a way to get somewhere useful without wrestling for control of the conversation.

The handoff mattered as much as the voice.

The work behind a useful conversation

The satisfaction at the other end

On September 9, 2026, PayNearMe announced that it had acquired Marr Labs’ technology assets and hired key employees. The founders joined a payments business serving more than 20,000 businesses and processing over $50 billion annually. It was another move from a specialist technology company into an established platform with customers and operational demands of its own.

In a conversation about the combination, Grannan described deployments handling more than two million calls a month. He stressed accuracy, reliability and the realities of regulated work. His stated ambition was grounded in the result: faster payment collection, lower costs and “customers who end the call satisfied.” After years of explaining what new technology might make possible, he was describing what the person at the other end should experience.

The objects along his career are easy to remember: early mobile software, a voice assistant, a camera covered in lenses. The connections are less visible and more durable. There are colleagues who returned, technical ideas that found different customers, and products tested by the inconvenience they were supposed to remove. For Grannan, the next conversation carries a familiar question. Has the machine actually made the task easier?