Ben Colman

Adding a second AI model doesn’t necessarily give you an independent check. If both models share the same blind spots, they can agree and still be wrong. For compliance work, the real test is whether an answer holds up against your source documents and evidence. This article from the Turbo AI Solutions team explains why that distinction matters.

1

Most founders see data. Nitsan Yerushalmi saw the human cost of slow payments. Some founders stumble into ideas. Nitsan Yerushalmi collided with hers head-on. Before anyone knew her name, before Monto existed, before she ever walked into The Zell Entrepreneurship Program at Reichman University, Nitsan was sitting deep inside the operational trenches of B2B businesses - working as a billing and collections specialist. And what she saw wasn’t theory. It wasn’t a market trend. It was pain. Invoices trapped in portals, payments stuck in limbo and hours wasted chasing money already earned. Systems that made smart operators feel helpless. She felt it so sharply that ignoring it stopped being an option. At the same time, her close friend Maya, coming from ERP implementation and startup operations, was watching the same chaos unfold from the systems side. And during their year together in Zell, the two of them made a decision most people never do: They stopped complaining about the problem, and started solving it. No technical founder. No perfect runway. No shortcuts. Just conviction. 𖧹 They signed their first customer during the program. 𖧹 They secured their first check by the end of 2021. 𖧹 And they officially launched Monto - built on lived experience, not guesswork. But here’s what most people don’t see: The Zell community didn’t end with graduation. It became the backbone of the journey, mentors who still show up, founders who still push each other, and a network that still sharpens the company every single week. And beneath all of it sits one belief that Nitsan still repeats today: “I want B2B payments to be as easy as tapping a card - no portals, no chasing, no friction.” That clarity is why Monto is now scaling into the U.S. market and actively partnering with industrial and enterprise companies who struggle with AP portals and slow payments. If you know leaders who live this pain (or operators who spend their week untangling payment chaos) send me a message. - - - - - - I’m happy to make the introduction. Ori Schnitzer Naomi Scheinman Liraz Sharabani Reichman University Reichman University Alumni Association American Friends of Reichman University (IDC) American Friends of Reichman University (IDC Herzliya)

85

9 Comments

I've interviewed some of the biggest names (founders & CEOs) across AI over the past few weeks and these are the latest discussions happening: 1. We are heading towards headless software 2. Security + privacy is the no.1 concern and priority 3. 3x the number of Partners leaving venture firms to go solo / setup themselves vs previous years 4. We should be owning our own intelligence 5. LLMs will need to payout creators for very high quality information & content 6. Vibe coding tools can copy ideas but they can't copy trust. Copy cats & founders are also on different paths 7. We don't need any new social networking sites 8. GPUs are lasting much longer than anticipated - in some cases ~8+ yrs 9. There is an opportunity to build a new version of Slack 10. Companies are staying private much longer vs going public because they now can.

4

While not surprising, this is sobering news. "63 percent of respondents said it’s acceptable to use AI tools without IT oversight if no company-approved option is provided." Governance never keeps up with the speed of technology, but this is a whole new level. https://lnkd.in/eh2eEGBW

29

5 Comments

Something to think about with your morning coffee ☕ Gemini AI Prompt: What if AI succeeds and OpenAI fails? Response Summary; This is a scenario often compared to the "Netscape Moment." ​In the 1990s, Netscape Navigator proved that the web (the technology) was revolutionary, but Netscape (the company) ultimately failed because the technology was integrated into the underlying infrastructure (Windows/Internet Explorer) or became a commodity. ​If AI succeeds while OpenAI fails, we are likely looking at a future where "intelligence" becomes a cheap, abundant utility rather than a premium product sold by a single vendor. Link to full response; https://lnkd.in/gbXRk3v7 #AI OpenAI #GeminiAI

4

2 Comments

BIG NEWS - Cisco Talos has discovered CLOSEDQUORUM, a new malware binary identified through its frontier AI tracking project, CAIRN. Disclosed on September 22, 2026, this discovery marks a Critical Milestone as the ** first reported autonomous AI command and control (C2) ** implant. ⚠️ ⚠️ Key Details of the Discovery -> Fully Autonomous C2: Unlike traditional malware that requires human operators to execute commands, CLOSEDQUORUM features **Completely Automated** command-and-control operations. -> The CAIRN Project: The malware was detected via Cisco Talos' CAIRN project, an initiative specifically designed for frontier tracking of AI-integrated malware threats. -> Shift in Cyber Warfare: Security researchers note that CLOSEDQUORUM represents a massive shift toward "effort displacement" for threat actors. By automating vast portions of the attack chain, malicious operators no longer need to be actively involved to sustain or adapt an active intrusion. ⚠️ ⚠️ The emergence of CLOSEDQUORUM confirms long-standing industry warnings that threat actors are shifting from simply using AI to write code to deploying fully autonomous, agentic AI weapons on the digital battlefield. #AICnC #ClosedQuorum #Hackers #CyberWarfare https://lnkd.in/gRXGk4-c

1

Conclusion Visa is executing a remarkably coherent strategy to position itself as the essential infrastructure layer for agentic commerce. The approach leverages existing competitive advantages (tokenization dominance, global scale, standards-setting authority) while adding new products (data tokens, spending controls, Trusted Agent Protocol) that address unique agentic requirements. The "shop for me" button is not just a catchy demo—it's the interface to a comprehensive platform designed to solve the trust, security, personalization, and control challenges that will determine whether autonomous AI agents can actually transact at scale. Most critically, Visa is moving fast while the window is open. With live transactions already flowing and general availability planned for 2025, the company is establishing potentially durable competitive advantages before the market structure solidifies. If the Trusted Agent Protocol gains wide adoption and Visa's APIs become the default integration for LLM platforms, the company could achieve in agentic commerce what it achieved in e-commerce and mobile: dominant infrastructure positioning that benefits from powerful network effects. The investment community should watch three key indicators: (1) merchant adoption metrics for Trusted Agent Protocol, (2) LLM platform launch announcements featuring Visa integration, and (3) issuer enablement progress. These will signal whether Visa is successfully executing this ambitious positioning strategy. Supporting Evidence "We're not just doing nice PowerPoints and presentations, but the really hard work to sort of weld it into the system of how agentic commerce will work going forward." — Oliver Jenkyn, V UBS Tech & AI Conference, December 2025 "We see considerable opportunity in agentic commerce... when we had the first wave of digital commerce with eCommerce, we set the standards, we led the product development and Visa was a significant beneficiary. Then you saw a second wave of commerce, which was mobile commerce. And again, Visa was the leader in terms of standards, in terms of product innovation... And now in this third wave of agentic commerce, we've been leading in terms of our role of setting the standards." — Ryan McInerney, V Q4 2025, October 2025 "I'm pleased to announce that we are now powering live agentic transactions and recently released a merchant agent toolkit to make it easy for developers to embed our solutions into workflows and agentic processes." — Ryan McInerney, V Q4 2025, October 2025 "We have more than 30 partners testing in our live sandbox, and we will soon enter the live transaction pilot phase with general availability to follow later this year as we see agentic commerce becoming a reality." — Ryan McInerney, V Q3 2025, July 2025 "Tokenization and authentication is at the very core of it. The basic principle is we're going to certify and authorize an agent, and we're going to take Will's payment credential, create a token and bin

2

THREAT LEVEL: Everyone assumes AI safety reports are PR theater. Anthropic's new 154-page threat intelligence report reads like the opposite of that. It covers eight months of disrupted misuse, December 2025 through August 2026, across seven categories: cyber operations, influence campaigns, surveillance, fraud, biological misuse, weapons development, and model distillation. A few cases that matter if you're building anywhere near DefenseTech or dual-use: a Russia-based actor used Claude Code to test kamikaze drone swarm logic. A state security-linked group ran a domestic surveillance operation that identified Hong Kong pro-democracy figures and Uyghur advocates as targets, and got shut down for it. A group tied to China's navy used Claude on anti-torpedo systems work. Alibaba allegedly ran 25,000 fake accounts trying to distill Claude's outputs straight into their own models. The abuse concentrated almost entirely on Haiku, Sonnet, and Opus -- the workhorse models running daily API volume. The frontier tier, Fable and Mythos, stayed nearly clean, showing up in exactly one distillation attempt. Every case in this report got caught and disrupted. Anthropic published the casebook anyway, fully aware it hands critics another news cycle about what AI enables. That takes actual confidence, backed by receipts. If you're building anything dual-use or defense-adjacent, this is what responsible looks like on paper. A documented detection-and-disruption track record beats a values statement every time a regulator or a prime asks the uncomfortable question. The founders who win the DoD and enterprise trust game over the next three years will be the ones who can produce something that reads like this report. Of course we have no idea how many bad actors got through without being detected.