Omer Har and the Value of Looking Outside

Omer Har offers an unusual explanation for leaving a research job: he thought he was better at managing people. He had worked at Microsoft Research in Cambridge, a place he remembers fondly. The attraction of a smaller company was the chance to lead sooner. Then comes the qualification, delivered with the instinct of someone who knows that employees have opinions too: ask the people he managed whether they agree.

It is a small joke with a useful complication. Here is a founder of a data company who will happily supply his own hypothesis, then suggest checking it against another dataset. His self-assessment comes with a request for independent verification. Somewhere, a performance-review form has found its natural author.

Today Har is the co-founder and CEO of Explorium. Its business is helping other businesses use information from beyond their own systems. The progression from researcher to executive has involved a similar willingness to look outside the familiar: outside a technical specialty, outside an original product plan, outside the records a company already owns.

The route through the laboratory

Har’s formal training began in computer science. He earned a bachelor’s degree at Ben-Gurion University of the Negev between 2004 and 2007, followed by a master’s between 2008 and 2011. Later, while his career was moving toward entrepreneurship, he studied for an MBA at Tel Aviv University from 2017 to 2019. The educational sequence makes room for both the algorithm and the organization that must pay for it.

His working life included software engineering at Intel and Microsoft, applied research at Microsoft, and data science leadership at Supersonic and ironSource. Those are different settings for the same awkward encounter: a technical solution eventually has to meet the world outside the technical team. An elegant model still needs useful information, a purpose, and someone prepared to act on its output.

In his early account of becoming a data scientist, Har described being drawn to difficult algorithmic problems while working as a software engineer. Microsoft Research offered applied machine learning at scale. Later, leading data teams in advertising technology brought another kind of discipline. A team could investigate a promising dataset, but the investigation had to justify the time it consumed.

That experience matters to Explorium’s origin. The company grew from a practical irritation familiar to its founders: useful outside information existed, yet discovering, acquiring and joining it to internal records could become a project of its own. A data scientist hired to improve a business could spend an expensive afternoon searching for the ingredients.

An expensive way to go looking

Explorium was founded in 2017 by Har, Maor Shlomo and Or Tamir. Shlomo supplied the original idea; Har brought experience of building models under commercial constraints. The early platform connected internal company information with outside data, then searched for useful features to improve predictions. The appeal was straightforward. A company’s own records describe its encounters with the world. They cannot contain everything happening around those encounters.

A sales database, for example, might record previous conversations with a prospective customer. It may say less about that customer’s changing business. An outside signal can add context. The technical difficulty is identifying the relevant information and connecting it to the correct entity, rather than merely accumulating a larger pile of records.

In 2020, Har explained the problem as one of uncertain returns. A team could not know beforehand which source would improve a model. It had to investigate, negotiate access and experiment. Even a talented group could end up choosing work it controlled over work that might produce a greater benefit. Explorium tried to make that search systematic.

By May 2021, Explorium had announced a $75 million Series C led by Insight Partners, bringing its announced funding to $127 million. Customers named at the time included PepsiCo, Melio and BlueVine. The uses ranged from lead scoring to demand forecasting. These were company milestones, achieved by a founding team and its employees. They also widened the arena in which Har’s technical judgment had to operate.

The CTO leaves his keyboard

By 2022, Har’s working day had changed. In the beginning, he wrote code. Growth brought management; a vice president of research and development later took on that responsibility. Har found himself working on strategic accounts, presales and sales engineering. For a technically trained founder, the interesting question became where his contribution would matter next.

Explorium was changing too. Customers showed more interest in accessible external data than in the platform’s sophisticated model-building capabilities. The company adjusted its emphasis. Its technology had offered several things; customer behavior helped identify which one deserved the foreground. Listening required a certain professional vanity to be put aside. Algorithms do not complain when they are given a smaller part in the brochure.

“By far the most important job any founders have is to find the right people.”

Omer Har, 2022

His account of a founder’s work is consequently less tidy than a ladder of promotions. Responsibilities change as other people arrive and customers reveal what they value. A title may persist while the job underneath it becomes almost unrecognizable. Har’s willingness to follow that movement helps explain the distance between the researcher and the executive now leading the company.

Start with the question worth answering

One principle has remained consistent in Har’s public advice: begin with the business problem. In 2022, he urged companies to design data work around a decision they actually needed to make. Which prospective customers were likely to convert? Which borrowers were likely to repay? Those questions should shape what gets collected, connected and used.

That is an unglamorous instruction in an industry with a generous appetite for collection. A company can acquire enough information to furnish a small library and still struggle to decide whom to call on Monday. Har’s approach puts the decision ahead of the inventory. The value of an additional field depends on what someone can do with it.

He also drew attention to small and medium-sized businesses, whose information can be difficult to assemble. They vary, change and disclose less than public companies. Understanding them requires more than assuming every useful fact arrives in a standard filing. For a supplier, lender or other business serving them, the missing context can be consequential.

A new customer with no patience for a dashboard

In March 2025, Explorium introduced AgentSource, a toolkit of business-data APIs for go-to-market AI agents, alongside an announced integration with Outreach. Har was its CEO. The customer’s task had changed again: software needed access to outside information while carrying out sales and marketing work.

The announcement described aggregating data from more than a hundred sources. Its central promise was to reduce the barriers between a developer and the information an agent required. A person might inspect several tools and reconcile their answers. An automated workflow needs those answers delivered in a form it can use as it goes.

Har’s own LinkedIn commentary takes aim at the practical cost of assembling multiple vendors: separate contracts, credit balances and interfaces. The point is commercial as well as technical. Every additional connection introduces another thing a builder must maintain. He is selling consolidation, and his argument rests on the accumulated nuisance of making disconnected products cooperate.

Another post described sixteen skills developed for Explorium’s own workflows, including contact enrichment and lead scoring. This puts the company in the position of using the kind of automation it supplies. The examples concern recurring business chores, where a pleasing demonstration has to survive another working day.

The leadership around him now includes Tamir as co-founder and chief operating officer, Lonny Merersky Varas as chief product officer and Amit Avraham as chief revenue officer. Those divisions of responsibility give a concrete shape to the founder’s changing job. A company needs product decisions and commercial decisions to meet somewhere. For Har, leading the organization means working across the specialties that once framed his own career, while other executives carry responsibilities he cannot sensibly perform alone.

Fresh facts, old argument

By the summer of 2026, articles carrying Har’s byline were examining the mechanics in detail. A July guide to business enrichment inside Codex separated identifying a company from adding fields to its record. Match confidence matters because a plausible answer attached to the wrong company can travel through a sales system with an undeserved air of authority.

An August guide turned to finding companies through events and intent. It distinguished learning more about a business already on a list from discovering a group of businesses that met new conditions. For an agent asked to find prospects, that difference determines whether the work can begin. A beautifully enriched list still contains only the companies someone already thought to put on it.

Later that month, a guide on data for automated sales development examined stale contacts, missing context and timing. It organized the problem into identity, context, timing and delivery. The argument carries Har’s earlier concern into a faster setting: when software acts repeatedly on old information, a small defect can become a recurring operational mistake.

The vocabulary has moved from predictive models to agents, enrichment endpoints and live queries. Har’s underlying preoccupation remains recognizable. Before making a decision, discover what the system is missing. Before automating that decision, make the missing information usable. His career has followed a comparable habit of revision: examine where the work is going, then change the job accordingly. The research instinct survives, even when the meeting calendar has claimed the keyboard.