Nir Laznik introduces himself in an order that few founder biographies preserve. Father first. Then information systems engineer. Then comes the company, the market, the acronym soup. In a long conversation about artificial intelligence and financial compliance, that sequence takes only a moment, but it tells you how he likes a system arranged: priorities declared, roles separated, important things placed where nobody can mistake them.
This is useful because Laznik’s current subject is a category built from blurred lines. Banks and fintechs are adding AI agents to calls, chats, email, marketing and customer service. Those agents promise volume and speed. They also create a question that grows less philosophical with every deployment: who checks what the machine says when the conversation becomes regulated?
Laznik’s answer is Sedric, the company he co-founded with data scientist Eyal Peleg in 2020. Sedric turns an institution’s policies into controls that can monitor communications and marketing, flag deviations, preserve evidence and support remediation. Its governing idea is almost old-fashioned in its neatness. A system doing the work should not be the only system deciding whether the work was done properly.
“Even the most compliance-oriented AI solution cannot just evaluate itself.”Nir Laznik, on independent AI oversight
The first company wore limited editions
The route to compliance ran through fashion, crowdfunding and Shanghai. In 2013, Laznik co-founded Out of X with Roee Lahav and designer Shany Elkin. The premise was visually lively and mechanically strict: independent designers uploaded prototypes, set a minimum production quantity and collected pre-orders. A piece went into production only after enough buyers committed. Each run stayed limited, often to tens or a few hundred garments. The name lived on the label: one item out of X.
The company initially tested demand in Western cities, including New York and London. Then a startup convention in Sydney rearranged the map. Investors there urged the founders to look at China. Interest arrived quickly. Out of X appeared on The Next Unicorn, a Chinese startup television program, entered Chinaccelerator and shifted its focus to Shanghai. Laznik lived there for several years as the company tried to connect Western independent designers with Chinese shoppers who wanted something less standardized than mass luxury.
It was the kind of founder story that photographs well: a young Israeli team, a giant market, a clever product, a flight east. It also ended. In 2017, after roughly three years, the founders shut Out of X. Laznik described the decision publicly as difficult. That post belongs in his biography because failure is not a decorative prelude to later funding. It is where the operator learns which parts of his confidence survive contact with an ending.
A less glamorous bottleneck
Sedric began with a different kind of excess. Financial products were multiplying across fintechs, traditional banks and embedded-finance companies. Rules were becoming more complicated, while customer communications scattered across voice, chat, email and social channels. Compliance teams often reviewed samples because they could not inspect everything. A two percent sample could be normal in call monitoring. The other 98 percent remained, in practical terms, a dark room.
Laznik and Peleg saw pressure collecting in the middle. Smaller financial institutions lacked the armies available to global banks. Larger banks had the staff but faced sprawling systems and a new generation of AI tools. Compliance, tasked with protecting customers and the institution, could stop an experiment it could not observe. Laznik’s bet was that visibility could change the answer from an anxious no to a defensible yes.
A $3.5 million seed round led by StageOne Ventures arrived in 2021. Three years later, Foundation Capital led an $18.5 million Series A, with Amex Ventures participating and existing investors joining. Sedric said the round brought total equity funding to $22 million and would expand its Tel Aviv AI lab and global go-to-market teams. At the time, the company reported fivefold revenue growth over twelve months and customers across the United States and Europe.
The numbers matter, but the product argument is more revealing. Laznik does not frame compliance as a robot replacement scheme. He frames it as a separate line of defense. Policies, jurisdictional differences and product details become a context layer. Communications flow through it. Exceptions surface. Humans remain accountable, but they receive a much wider view than a random sample and a paper trail that can be examined later.
His advice for introducing that machinery is notably uncinematic. Start with the data already available. Put it into existing processes. Let teams develop an appetite for better evidence, then widen the system. The sequence avoids the grand unveiling that leaves operators staring at a new dashboard and wondering which meeting it belongs in. Laznik has also stressed practical requirements such as reducing personally identifiable information, supporting data-residency choices and encrypting information from end to end. These are the plumbing decisions on which an AI promise either becomes institutional practice or remains a pilot with excellent slides. In regulated finance, the dull details are often the product.
The eye in the sky needs its own job
The cleanest version of Laznik’s thinking appears when he talks about AI agents. A vendor might build a voice bot, then add another model to judge the first. Yet both may live inside the same product and share the same incentives, assumptions or blind spots. Laznik borrows the logic long used for human collectors: experience does not exempt anyone from monitoring. An AI agent should not receive a special dispensation simply because its prompt contains the policy.
Consumers make this especially important. A designer can specify how an agent should open a call, disclose terms and handle familiar objections. The consumer supplies the unscripted half. Questions wander. Edge cases arrive without appointments. A model that behaves impeccably in a lab can enter unfamiliar territory in production. Independent oversight creates somewhere for those moments to become visible.
“There is a very common bias and misconception about comparing AI to perfection, where suddenly people forget that you need to compare AI to the current state.”Nir Laznik, on realistic benchmarks
This is Laznik at his most pragmatic. He is not arguing that machines deserve leniency. He is asking for an honest benchmark. Human review is partial. Existing processes miss things. The useful comparison is between two operating systems with measurable strengths and failures. Perfection makes a fine aspiration and a terrible control group.
Sedric faced a related choice. If its models could spot a problem in marketing copy, why not rewrite the copy too? Laznik has said the company chose to remain on the oversight side. Creation and judgment would stay apart. The decision sacrifices a tempting expansion in exchange for a clearer role. In a sector fond of platforms that promise to do everything, saying what your product should not do can be a quiet form of product discipline.
The long way to a simple rule
In 2025, Sedric added a strategic venture loan from HSBC Innovation Banking. In 2026, it won the Banking Tech Awards USA category for communications compliance and joined the American Fintech Council. Its vocabulary has widened from monitoring to “agentic compliance,” a model in which AI systems execute parts of the compliance workflow under professional supervision. The ambition is to break the old equation where every increase in business volume requires a matching increase in compliance headcount.
There is a risk that any phrase containing “agentic” will age with the speed of conference lanyards. Laznik’s underlying rule is sturdier: codify the policy, separate execution from review, keep the evidence, and give accountable people enough visibility to act. Those are not glamorous verbs. They are the verbs that remain after the demo ends.
His two-company career now forms an unusual pair. Out of X tried to make scarce fashion accessible by coordinating designers, demand and production across borders. Sedric tries to make abundant financial communications governable by coordinating policies, models and review. One dealt in limited editions; the other deals with virtually unlimited content. Both required Laznik to enter a market whose complexity could not be managed from a comfortable distance.
The father-first introduction returns here. It is not evidence of a management doctrine, and it need not carry that burden. It is simply a useful clue about how Laznik orders a sentence. The personal role comes before the technical credential. The technical credential comes before the company pitch. In his work, too, responsibility precedes capability. The machine may speak, sell, summarize and assist. Someone still has to watch.