Hey — It’s Nico.

Welcome to another Failory edition. This issue takes 5 minutes to read.

If you only have one, here are the 5 most important things:

• Skybox Security, an Isreali cybersecurity startup, has shut down — learn why below.

• How to Design an Org for Founder Mode.

• Claude 3.7 is out and its really good.

• AI search engine Genspark raised $100M to compete with Google and Perplexity.

• Evo 2 is a new AI model that can design biological life — learn how below.

Let’s get into it.

Unlock Free Access to 200M+ Prospects! ^AD

Redrob helps startups connect effortlessly with top prospects, talent, and investors.

Use natural language search across 200M+ profiles—and put outreach on autopilot to collect high-value meetings.

Include video for a personal touch–or offer cash incentives to cut through the noise and entice your top leads.

AI so easy your parents could use it 😌

Try it free for up to 6 months! -- No Credit Card, No Catch.

This Week In Startups

🔗 Resources

The end of programming as we know it.

Non-obvious SEO advice for startups.

How to Design an Org for Founder Mode.

How to get small business ideas and evaluate them.

📰 News

Claude 3.7 is out and its really good.

Apptronik’s humanoid robots take the first steps toward building themselves.

Thinking Machines Lab, Mira Murati’s new startup is looking to raise $1B to join the AI race.

Crunchbase’s AI can predict startup success with 95% accuracy.

💸 Fundraising

Quantum Machines raises $170M to make quantum computers.

AI search engine Genspark raised $100M to compete with Google and Perplexity.

Mercor, an AI recruiting startup, raised $100M.

Nomagic, a Polish AI robotics, raised $44M.

Fail(St)ory

Fatal Error

A few days ago, Skybox Security, an Israeli cybersecurity startup, shut down, laying off all 300 of its employees. The company had been around for more than two decades, serving hundreds of major enterprises worldwide, but in the end, it couldn’t sustain itself.

What Was Skybox Security:

Founded in 2002, Skybox Security developed tools to help organizations analyze their firewalls and test overall network security. Over time, it expanded its offerings, but its transition to SaaS and newer product lines struggled to gain traction.

For years, Skybox relied on its existing customer base, but as those relationships faded, it failed to attract enough new business to keep growing. While its technology was respected, its solutions never became must-have products in the cybersecurity space.

Despite generating over $500 million in total revenue over its lifetime and serving major corporations across 50+ countries, Skybox couldn't maintain momentum.

The Numbers:

• 📅 Founded in 2002

• 🏢 Served hundreds of Global 2000 companies across 50+ countries

• 💰 Raised a total of $288M, including a massive $150M round in 2017

• 🔥 Revenue exceeded $500M over its lifetime

• 📉 Laid off 300 employees upon shutting down

Reasons for Failure: 

• Stagnation and Failed Expansion: Skybox built a solid foundation with its initial firewall and network security analysis tools, but its efforts to expand beyond these core offerings didn’t take off. The company's shift to SaaS and newer product lines failed to gain widespread traction, making it overly dependent on a shrinking customer base.

• Late-Stage Funding Pitfalls – In 2017, Skybox raised $150 million in a funding round led by CVC Capital Partners and Pantheon Ventures, bringing its total investment to $288 million. Instead of accelerating growth, this influx of capital seemed to highlight deeper problems. Skybox couldn’t scale fast enough to justify the funding, and growth stagnated.

• Leadership Transition Issues – Founder Gidi Cohen led the company for 20 years before stepping down in 2022, handing the reins to Mordecai Rosen. The leadership change didn’t reverse the company's trajectory, and Cohen himself stated that he had been entirely disconnected from operations for the past two years. This disconnect may have contributed to Skybox’s inability to correct course in time.

• Market Positioning and Relevance – Skybox combined firewall policy management with vulnerability management, but according to analysts, this combination never gained widespread market traction. The company found itself stuck between two segments without dominating either, making it vulnerable to shifts in cybersecurity trends.

Why It Matters: 

• Skybox is a case study in why sticking to legacy customers without innovation can be dangerous for a tech company.

• Leadership transitions can be critical turning points—when a founder leaves, companies need a clear strategy to maintain momentum.

• Even well-funded startups in established industries can fail if they struggle to differentiate themselves in a competitive market.

Trend

Evo 2

I know, I know—I talk about AI every week. But stick with me, because this one’s different. Most AI models I talk about focus on text, images, or programming, but what if AI could be used to understand, predict, and even design biological life?

Enter Evo 2, a new AI model that doesn’t just generate words—it generates DNA sequences.

Why It Matters

• AI Beyond Text and Code: Evo 2 moves AI beyond chatbots and coding assistants into genomic science, paving the way for AI-driven bioengineering.

• Massive Implications for Medicine and Biotech: With the ability to model and predict genetic mutations, Evo 2 could revolutionize disease research and drug development.

• An Open-Source Approach to Biology: Unlike many proprietary AI models, Evo 2’s creators are making it open-source, accelerating innovation across biotech industries.

• Bridging AI and Life Sciences: Entrepreneurs who think AI is just about automation might need to reconsider—biology is becoming the next frontier for AI applications.

What is Evo 2?

Developed by Arc Institute, Evo 2 is a foundation AI model built specifically for genomic modeling, prediction, and DNA sequence generation. While most AI models work with text or images, Evo 2 was trained on 9.3 trillion DNA base pairs—essentially learning the “language” of life itself.

With 7B and 40B parameter versions and a 1 million-token context window, Evo 2 can analyze long genetic sequences, predicting the effects of genetic changes at an incredibly detailed level. In simple terms, it can help scientists understand which DNA mutations might lead to diseases—like identifying small tweaks in genetic code that could increase the risk of a certain illnesses like breast cancer.

Arc Institute is taking an open-source approach, making the model weights, training code, inference code, and a dataset (OpenGenome2) available on GitHub. They’ve also developed tools like Evo Designer, which lets researchers create and edit DNA sequences, potentially speeding up breakthroughs in genetic research. And to prevent any risky applications, Evo 2 was trained without virus-related DNA, ensuring it can’t be misused to modify harmful pathogens.

Interestingly, Greg Brockman, OpenAI’s co-founder, contributed to the project during his sabbatical, bringing his expertise in AI infrastructure to improve Evo 2’s data processing capabilities.

AI Beyond LLMs

Evo 2 represents a major shift in how AI is applied outside of traditional tech fields. Here’s what that could mean:

• Bioengineering Gets a Boost: Evo 2 could help design synthetic genomes, optimizing DNA for research, agriculture, and even biofuel production.

• New Frontiers in Medicine: With its predictive capabilities, Evo 2 could accelerate the development of personalized medicine, identifying which genetic variations contribute to diseases and how to treat them.

• Startups in AI-Driven Biotech: This model lowers the barrier for startups looking to innovate in genetic research, providing open-source tools for rapid experimentation.

But most importantly, Evo 2 is a reminder that AI goes way beyond LLMs and chatbots. While language models like GPT and Claude dominate the conversation, it’s important to remember that AI is also making waves in niche industries:

• AlphaFold, developed by DeepMind, is revolutionizing drug discovery by accurately predicting protein structures.

• PathAI is enhancing cancer diagnostics by analyzing medical images with unprecedented precision.

• Covariant AI is enabling warehouse automation with adaptable robotic systems.

Help Me Improve Failory

That's all of this edition.

Cheers,

Nico