David Sosna Keeps Asking What Happens Next

A late order for an important customer is a small business disaster with an especially irritating feature: somewhere, a computer probably knew about it. The information existed. The person who needed it simply had not been told. David Sosna was thinking about that gap long before an AI assistant became something you could summon from a browser tab.

In the 1990s, his company Gilon worked on databases and business intelligence. Employees analyzed information held in corporate data warehouses and made it visible to managers. Then came a more interesting possibility. Could the database notice a problem as it happened and alert somebody who could do something about it? A beautiful report delivered after the damage had been done was still a report delivered after the damage.

That question helped produce Actimize. Later, the same appetite for timely information would take Sosna into everyday banking through Personetics. Today, as founder and CEO of Sympera AI, he is working on the relationship banker's crowded day. The companies have different customers within the bank, different products and different ambitions. Read together, they suggest a career spent shortening the distance between knowing something and acting on it.

The database learns to interrupt

Sosna graduated from Ben-Gurion University's Department of Industrial Engineering and Management. His professional starting point was the machinery of information: collecting it, organizing it and putting it to work. He founded Gilon in 1994. The next venture grew from an awkward realization inside that business. Companies had invested heavily in databases, yet the systems were better at supporting management analysis than keeping up with the millions of operations happening around them.

Actimize was founded in 1999 by Sosna, David Govrin, Guy Greenberg and Boaz Pe'er. Greenberg had been Sosna's partner at Gilon. Govrin would become a returning character in Sosna's working life. This was a company formed around a practical absence. The team wanted tools that could analyze events continuously and get an appropriate message to the right employee or system. When they looked for those tools, they could not find what they wanted.

“We therefore decided to found a company to develop the necessary product,” Sosna said in 2001. There is a pleasantly workmanlike quality to that explanation. No thunderbolt is required. A customer has a problem; the available equipment is inadequate; a group of people builds something.

The financial sector became an important home for that approach. Actimize developed software for fraud prevention, anti-money laundering and compliance. In July 2007, NICE announced an agreement to acquire it for approximately $280 million. Its announcement said Actimize's systems were analyzing close to a billion transactions daily. Sosna would remain CEO of the business under NICE. A question about information arriving in time had become part of the plumbing of institutional finance.

The same transactions, a different beneficiary

The leap to Personetics involved a change of perspective. A transaction could indicate danger to a bank. It could also reveal something useful to the person who made it. Sosna and Govrin returned to that familiar material with a different recipient in mind: the customer opening a banking app.

A May 2011 funding announcement described Personetics as founded in 2010 and disclosed $6.5 million from Carmel Ventures and Sequoia Capital. Its early Digital Banker product was designed for web, mobile and text banking. The ambition was to handle more customer interactions within those channels, saving people a trip to the call center. Even at the start, the product depended on understanding a customer's circumstances rather than displaying the same menu to everyone.

By 2016, Sosna could explain the continuity plainly. Much of the founding team had worked together at Actimize. They already understood bank transaction data and the technical requirements of analyzing it. The knowledge acquired while looking for fraud could be applied to serving customers. He was equally clear that familiarity did not settle everything: “Obviously we still had to learn a lot and we are still learning.”

Banking applications tend to organize the world into accounts and products. People generally arrive with a more immediate concern. They want to understand a change, complete a task or decide what to do next. Personetics worked within that mismatch. The attraction of a transaction was the context it supplied. With enough context, a bank could make a useful suggestion at a moment when the customer might welcome one.

“Obviously we still had to learn a lot and we are still learning.”

David Sosna, 2016

A useful bank should have good timing

Sosna's description of the product in 2020 made adoption sound almost domestic. It belonged inside the banking app people already used. Customers should not have to install another application or begin by feeding a new service their data. The bank already held transaction information. The work was to turn that information into something personally relevant without giving the customer another small administrative job.

He described increasing levels of capability: organize the data, identify what matters, offer advice and eventually automate an appropriate action. This was the logic behind Personetics' self-driving finance proposition. Moving money automatically was a later step in a progression that began with making account activity intelligible. The sequence matters because it makes the ambition inspectable. Each step has to earn its place in an ordinary person's day.

The partnerships give that ambition some texture. At Belgium's KBC, a published Personetics case study reported that one in three enrolled users actively engaged with the solution. Insights received an average rating of 4.65 out of five. Those were company case-study results for a particular deployment, rather than a promise about every bank. They nevertheless offered a more useful test than whether the underlying software sounded clever.

In Southeast Asia, UOB and Personetics expanded their partnership with an automated savings feature on the TMRW app. Their 2022 announcement said the collaboration, which began in 2018, had delivered more than 150 million personalized insights across Indonesia, Malaysia, Singapore and Thailand. Here was the original idea traveling: transaction information could become a nudge, and a nudge could become an action.

Thirteen years, then another beginning

Running that operation also meant building a company large enough to serve banks in different markets. In 2020, Sosna described a team of more than 200, with over 70 people hired during 2019. Keeping distributed employees talking regularly and meeting periodically was part of his answer to the growth problem. He also discussed recruitment, inclusion and local volunteering. The software needed coordination; so did the people making it.

In February 2024, Personetics announced that Udi Ziv would succeed him as CEO. The company described a 13-year period under its co-founder's leadership and reported more than 130 financial institutions, 35 markets and a reach of 135 million customers. Sosna welcomed Ziv's experience in enterprise software. The handover came with a large installed base and the organizational demands that go with it.

His next company returned to the place where an unanswered question becomes a missed opportunity. Sympera AI focuses on relationship banking, especially the bankers serving small and medium-sized businesses. These bankers may have hundreds of customers assigned to them. A portfolio can contain more worthwhile conversations than a working week can accommodate. Knowing every client's name does not create another afternoon.

Who gets the next phone call?

Sympera's proposition is to help bankers decide where to spend their attention. It combines information about businesses with what the bank knows, then identifies opportunities and suggests action. CTO Daniel Gordon brings AI research experience to the team. The intended result is a better prepared human conversation: a banker reaching the appropriate business with a relevant reason to talk.

In March 2025, Sympera announced a $10 million seed round co-led by Nyca Partners and Viola Ventures. There was continuity among the backers, too. Viola's Avi Zeev said he had partnered with Sosna for more than 25 years and that this was the fourth company of his in which he had invested. Repeat entrepreneurship, in this case, comes with people who have watched earlier chapters at close range.

Sosna's public argument for AI leaves plenty of work for bankers. Writing in July 2025, he challenged the idea that the technology's main value lay in cutting costs or replacing human insight. He also resisted treating generative AI as the answer to every problem. “GenAI is just one form of the technology,” he wrote. His preferred discussion includes prediction, pattern detection and systems that can coordinate tasks. A banker still has to earn the client's trust.

By May 2026, Sympera reported that Republic Bank of Chicago was expanding its use of the platform after a pilot. The bank's business banking teams would receive suggested actions inside Microsoft Dynamics. External signals such as hiring activity and company changes would be combined with internal relationship information. That is a concrete place to judge the idea: the working environment a banker already opens, where another piece of software must justify the attention it consumes.

The short list is the product

In an August 2026 conversation on Terminal Value, Sosna returned to the difficulty of fitting AI into actual work. The discussion covered narrowing a large pool to a handful of researched opportunities, managing the cost of different models and keeping outreach subject to human review. It also addressed the hard part that a demonstration can politely omit: getting people to change how they work.

There is a particular impatience running through his public comments. Adding another tool can leave the original burden intact if the banker still has to ask every question, assemble every clue and decide where to begin. His concern is with the process surrounding the technology. An assistant waiting to be consulted has a different job from a system that has already done useful preparatory work before the banker arrives.

The early Gilon problem concerned an employee who needed to hear about a late order. Actimize brought that urgency to financial risk. Personetics carried it into the customer's banking experience. Sympera now applies it to the person deciding which customer to call. Sosna has changed companies repeatedly while continuing to work on the same stubborn interval: the time between an event and a useful response. The next test is whether a banker, looking at tomorrow's diary, finds a better conversation waiting there.