When AI makes the call, who takes the blame? Inside the company that says it has the answer

You've probably spoken to AI this week without even realizing it- when you called your health insurer, your bank, or followed up on a test result. Now imagine that happening across every hospital, bank, and defense organization in the world. AI isn't coming. It's already here. And the real question is: when it gets something wrong, who's responsible?

That's the question Tel Aviv-based JEEN AI says it was built to answer.

A different category of AI

JEEN doesn't build AI models. It builds the guardrails around them, so that when a large organization puts AI to work on something serious, the answers can be trusted, tracked, and, if needed, traced back to a human decision.

"Many organizations need something different," says Moti Krispil, JEEN's Chief Strategy Officer. "JEEN provides the foundation for mission-critical AI. Mission-critical AI is when making the right choice — or making it wrong — may have a detrimental impact."

Krispil draws a sharp line between two very different uses of AI. Using ChatGPT to draft an email is one thing. Using AI to make medical, financial, or operational decisions, where a mistake has real consequences, is another problem entirely.

That is the problem JEEN says it solves. Its platform sits on top of an organization's AI systems and adds three things: security controls so sensitive data never leaks out, verification so an AI's answer is checked and grounded before anyone acts on it, and a built-in point where a human expert signs off before anything final happens.

Its typical clients are the kinds of organizations where a wrong answer isn't an inconvenience but a real problem: banks, insurers, hospitals, telecom companies, and defense and government bodies.

The problem hiding inside large organizations

Krispil's pitch to CEOs starts with a warning. Fast forward 12 months, he tells them, and an organization of 10,000, 20,000, or even 50,000 employees will have thousands of AI tools and agents running with no oversight. Nobody is tracking what they cost. Nobody knows who is responsible when one of them gets something wrong.

This is already happening inside organizations most people would assume have it under control. Employees are using AI tools — sometimes dozens of different ones — with zero oversight, feeding sensitive data, patient records and financial information into systems nobody at the top has authorized, or even knows exist.

"Most individuals have no clue how exposed they are, from a personal and professional basis," Krispil says. "Enterprises understand the risks, but the pace of technological adoption outpaces their ability to guardrail it."

Inside Maccabi: cutting a 30-minute case to 30 seconds

The clearest example of JEEN's model in action is Maccabi Healthcare Services, one of Israel's largest healthcare providers and insurers, with 2.6 million members. Every day, thousands of patient inquiries come in involving reimbursements, coverage questions, and medical rights. Handling each one used to require an expert, sometimes a physician, to manually work through medical records, past call transcripts, and documents just to decide how to proceed.

JEEN's system now digests the medical records, approvals and call transcriptions automatically, producing a distilled case summary with a recommendation, a confidence level, and the supporting references. That summary is presented to the physician in a simple app. The doctor reviews it, approves or overrides it, and a letter goes out.

The result: average handling time has dropped from 30 minutes to as little as 30 seconds, with a self-service success rate near 85%, according to Krispil. But he frames the win less as a productivity story and more as a human one. Israeli doctors often have only seven to ten minutes per patient, and time reclaimed from paperwork is time spent actually looking at the patient and hearing them out.

Crucially, a human always makes the final call. That human-in-the-loop principle, Krispil said, is non-negotiable across every sector JEEN operates in- healthcare, finance, and defense alike.

Where the internet isn't allowed

Defense and intelligence work raises a different problem: not just accuracy, but security. Every query sent to an AI system outside an organization's own walls is data leaving the building, which is a serious liability in military and intelligence work.

Krispil describes one use case in broad strokes. During an unfolding emergency, thousands of sensors, video feeds, phone calls, maps, and documents come in at once, and each piece on its own tells you very little. JEEN's system connects the dots across all of it, surfacing patterns before a human analyst could spot them and giving decision makers a recommendation fast enough to act on.

For its most sensitive deployments, JEEN runs entirely inside a client's own network, with no internet connection and nothing leaving the building. Oded Tahori, CEO and Founder of JEEN AI, says the goal is to give an organization the ability to manage its own private version of a tool like OpenAI internally, with the same generative power but walled off completely from the outside world.

Who's actually using it?

JEEN's client list spans several of Israel's most critical sectors. In financial services and insurance, that includes Isracard, Direct Insurance and Bank Mizrahi. In energy and industry, Delek and the Israel Electric Corporation. In education and the public sector, Reichman University and Ben-Gurion University. On the defense and government side, JEEN works with sovereign and mission-critical defense agencies it cannot name publicly.

As for cost, it's worth a brief mention. AI usage inside large organizations can rack up bills fast if it isn't monitored. There have been cases where a single employee's unmonitored AI process has generated a bill in the tens of thousands of dollars in just a few hours. This is part of why JEEN builds spending controls into its platform alongside its security and accuracy guardrails.

From 30 people to a global company in two years

JEEN has grown fast. Tahori says the company has roughly quadrupled headcount in the past year alone, from around 30 people to 130 in Israel plus another 20 across Singapore, London and New York- including a defense-focused team drawing on backgrounds from Israeli intelligence and security services.

In under two years, the company has raised $25 million, signed contracts worth tens of millions, and expanded well beyond Israel, with customers now in North America, Latin America, Asia-Pacific and Europe.

Tahori, notably, took JEEN public on the Israeli stock exchange early — an unusual move for a startup, but one he says reassures the enterprise and defense clients who want stability from a vendor handling their most sensitive systems.

"Mission critical AI" as the next cybersecurity

Both Krispil and Tahori argue that mission-critical AI is shaping up to be its own category- the way cybersecurity did before it. Israel became a cybersecurity powerhouse, the argument goes, because its innovation was field-tested, sometimes battlefield-tested, at scale. JEEN is betting the same playbook applies here.

"Imagine AI running for days autonomously," Krispil says. "Companies must have these guardrails, and they must have this infrastructure, to make it safe, secure and confident."

But when asked what keeps him up at night, Krispil is clear he isn't a doomsayer. "What keeps me awake at night is AI being used for the wrong purposes, handled by the wrong entities," he says. "I believe the ability to instill confidence in a safe and secure way of using AI is a good place to start."