Danny Lieberman @flaskdata

@BeingTheWriter cool!

Join me on Life Sciences Today I’m looking for 10 cutting edge clinical AI CEOs to join me on Life Sciences Today. This is your opportunity to create a new category of value in the US healthcare market - your category - not Epic’s. We are part of the Healthcare IT network with > 120,000 followers. The show is on fire with 60+ shows in 12 months with guests from Sanofi, Flatiron Health, Biorce and Lindus Health - the anti-CRO. You can pitch me to appear on the pod with a DM. You can see Life Science Today here on Youtube, Apple, Spotify, Google and all your favorite podcast channels. I write a weekly essay on strategy, technology, and decision-making. I am founder of OpenCRO and advise WHO EU on cyber and privacy in digital health. youtube.com/playlist?list=…

@MartinGTobias and it's always the guys with the worst product who try to retain you in the worst way

@MartinGTobias And anti-patterns help survival like OutOfCash newsletter.dannylieberman.com/p/the-outofcas…

@BeingTheWriter best to send me an email - dl [at] software [dot] co [dot] il

@BeingTheWriter Just DM me

@BeingTheWriter Sure!

How much would you pay for organizational memory - of your toughest competitors

The 11 biggest one-person clinical AI businesses in 2026 Everyone wants to build the next Epic killer. Meanwhile, solo operators are quietly building $300k–$2M businesses by attacking the boring bottlenecks nobody wants to touch. 1. The FDA whisperer Cybersecurity submissions for connected medical devices. Every connected device now needs one. Most founders have no idea how to write them. 2. The procurement unlocker Vendor security reviews that shorten hospital sales cycles from 18 months to 6. 3. The protocol auditor Protocol consistency audits before first patient in. Catch one hidden inconsistency early and save millions downstream. 4. The drift watcher Monitoring workflow changes after AI deployment. Everyone sells deployment. Nobody owns day 90. 5. The reimbursement copilot Helping founders understand payment pathways before commercialization. 6. The SBOM ninja Open source and SBOM risk analysis for regulated systems. Boring work. Huge moat. 7. The discrepancy hunter Finding clinical trial data issues before database lock. 8. The nursing friction killer Removing operational drag from care teams without replacing clinicians. 9. The validation packager Submission-ready AI validation evidence for regulated AI systems. 10. The agent containment expert Permissions, containment, and audit trails for clinical AI agents. 11. The knowledge operationalizer Turning 30 years of expert healthcare knowledge into operational systems. Most people think healthcare AI is about replacing clinicians. It’s not. The money is in the friction: the submissions, audits, handoffs, procurement reviews, and operational gaps between the brilliant idea and the patient. Boring bottlenecks are the highest-margin businesses in healthcare.

@JoelWBerry What a shocking thought - Jews refusing to go to the gas chambers again,

@LauraLoomer “Repeat a lie often enough and it becomes the truth”, is a law of propaganda often attributed to the Nazi Joseph Goebbels. Among psychologists something like this known as the "illusion of truth" effect. Guess this makes these guys colleagues of Nazis.

@thesohom2 Right now - doesn't look like the market is slowing down but tbh - it's hard not to feel the 2002 deja vu

COVID grease fire + GenAI hype cycle = déjà vu The COVID bubble was a sudden accelerant — free money, no diligence, insane valuations. The GenAI hype is the accelerant’s sequel — this time with GPUs instead of government checks. The parallels to the dot-com era are uncanny: Dot-com 1999 → Raise big, fast, because “the Internet changes everything.” GenAI 2023 → Raise big, fast, because “LLMs change everything.” Both cycles sucked in huge amounts of capital with famously thin diligence. Both created frothy valuations far ahead of adoption curves. And in both, most of the winners will look nothing like the companies that raised the biggest rounds. The dangerous overlap right now? COVID money extended the life of companies that should have shut down earlier… just long enough for them to pivot to “we’re an AI company now” to chase the next capital wave. The survivors in Q4 2025 won’t be “AI companies.” They’ll be real businesses that: Use AI to solve real, painful, paid problems. Fit into workflows customers already use. Don’t need GPUs or hype to stay alive. The dot-com bust didn’t kill the internet — it cleared the way for companies that could actually execute. The AI bust will do the same.

AI agents aren’t risky. Bad system design is risky. I keep seeing the same failure patterns: • weak access control • weak isolation • unbounded execution • no system boundary Then people call it “AI safety” or “AI governance.” It’s neither. It’s what happens when you give any process too much power and no containment. Execution controls help. But they sit on top of fundamentals: → least privilege → isolation → resource limits If your system can’t contain failure, you don’t have safety. You have exposure. We map these as full threat paths — how systems move from “allowed” to “dangerous” — and turn them into prioritized countermeasures. Comment “framework” and I’ll share it.

@EddiYesilo23954 Hi - just saw your DM on X sounds pretty cool to use Reddit and it makes total sense I just launched a new service as software - OpenCRO - FDA cybersecurity threat modeling for MedTech.

In Jan 2024 I unretired at age 73. With AI we are all unretired. This new chapter is about using our experience and wisdom to enrich our lives and the people around us. And make some money on the way. DM me and I'll show you how I did it.

I left Kit a year ago. Their UI is despicable and their online support uses an ai chat bot trained on incorrect data. Yeah - get close.com if you need landing pages use ChatGPT, Grok or Claude code I built the landing page for OpenCRO in 15' using Warp Code and Claude opencro.com

@rispectrum Respect. you're doing a good job. Thanks for sharing Here's the OpenCRO risk framework for software security. dannylieberman.com/downloads/Busi… The 7-step risk analysis loop and prioritized, cost-effective countermeasure plans built for regulated software.

x.com/i/article/1848…

@outreach_nicola what are we going to do about it @outreach_nicola?

מהמקפצה: מערכת המשפט מגדילה את ההימור בידיעה שאיש לא יעצור אותה makorrishon.co.il/news/local/art…

@rispectrum what kind of architecture do you like for a deterministic work flow like that? how do you do a software security assessment?

@ethereumdegen @AggieCapitalist 😂