“AI is so incredible. This technology is going to change the world.”
That sentence hits hard when it comes from someone who actually treats it as a responsibility, not a slogan. Joseph F. Kraus, co‑founder and CEO of Radical AI, looks at the current AI startup boom and sees a problem: most companies are solving easy, small things while the real hard problems sit untouched.
Plenty of founders are building tools to shave a few minutes off workflows or add 2 percent efficiency to a software process. Joseph and his co-founders chose a different path. Their focus is artificial general intelligence for scientific discovery, starting with materials science, in order to disrupt a scientific process that has been mostly unchanged for about 150 years.
As Joseph puts it, we are already good at “one, two or five percent improvements.” What we are bad at is novel discovery, which is exactly where the biggest human challenges live.
In this story, we will walk through how Radical AI came to be, why materials science is their beachhead, how they raised a huge pre-seed round in minutes, and what kind of culture you need if you want to build a mission-driven AI startup instead of yet another tiny tool.
Why Most AI Startups Aim Too Low
Joseph’s frustration is simple: AI can help cure disease, transform energy, and unlock new materials that make entire industries safer and cleaner. Yet most startups are choosing the equivalent of picking coins off the sidewalk.
At Radical AI, the starting assumption is different:
- The hardest problems are the most important problems.
- If you solve them, you do not just build a good company, you “fundamentally reshape human trajectory.”
- That kind of work will always be uncomfortable and slow at the start.
Instead of asking “What can I build fast and raise money for?”, the question driving Radical AI is closer to, “What is broken at the foundation of how we do science, and how can AI help fix it?”
To understand why that matters, you first have to see how slow materials science is today.
Radical AI’s Edge: 370x Faster Than Human-Only Research
New materials sit at the root of almost every big technical shift: cleaner batteries, lighter aircraft, safer armor, faster chips. But the process of discovering and scaling a new material is painfully slow.
- Typical materials discovery timeline: 10+ years
- Cost: huge, with most ideas never reaching real products
Radical AI is building AI agents that:
- Read and index millions of scientific papers
- Operate at about 370 times the speed of a human scientist
- Learn what a field has already tried, so they do not repeat old dead ends
In practice, that means a system that can:
- Read millions of publications
- Simulate billions of potential materials
- Suggest and help test thousands of candidates in a tight loop
Instead of a human scientist drowning in PDFs and spreadsheets, you get an AI that handles the indexing and pattern finding, while the human focuses on judgment, questions, and direction.
A Mission That Started at Home: “Be the Best in the World at What You Love”
Long before Radical AI was a startup idea, Joseph grew up with a simple but demanding message from his dad: do something you truly love, and aim to be the best in the world at it.
It was not about chasing status. It was about deep engagement. Being so absorbed in the work that you forget to check the clock. That mindset left Joseph with a quiet but constant question in the back of his mind:
Where can I make the biggest impact with the skills I have or can build?
That question pushed him toward science, toward service, and eventually toward entrepreneurship. It also became the lens he used to judge whether a path felt meaningful enough to pursue for decades, not just a few years.
Learning Mission First in the National Guard
As an undergraduate, Joseph met someone in the US National Guard and was drawn to the idea of serving while still in school. He had dreamed of going to a US military academy, saw the prestige, but also saw the deeper appeal: shared mission, clear values, and real responsibility.
The National Guard structure made sense for him:
- One weekend of drill each month
- Two weeks of training each summer
- Ability to keep studying while serving
He enlisted in his junior year.
Basic training became a crash course in mission-driven teamwork. He stood beside people from New York, Georgia, Maine, the Midwest, California, Silicon Valley, San Diego, and everywhere in between. Different backgrounds, different beliefs, one clear purpose.
Everyone was there to help build a stronger military to protect the freedoms the US talks about around the world. That shared purpose came with a big internal shift: he had to “remove myself from the equation” and put the team mission first.
Those months planted a seed. The feeling of working inside a strong mission did not fade when he went back to school. It followed him into every big decision that came after.
Why Rice University: Go Where People Work on Big Problems
After basic training, Joseph returned to finish his senior year, then headed off to more training, and eventually to graduate school at Rice University.
Rice was not an accident. He had presented his undergraduate research at a symposium, won best presentation, and as a result, sat down with faculty and grad students to talk about their work.
A pattern stood out. They were not doing science just for papers. They were pushing on problems that could be transitioned into real use in the world. That mix of deep science plus real-world targets matched what Joseph cared about.
So he started grad school at Rice while still serving in the National Guard. He did not yet know if he wanted to be a professional scientist, a patent lawyer, or something else. But he kept coming back to the same question: where could he have the most meaningful impact?
From Army Research Lab Frustration to Startup Thinking
While at Rice, Joseph also worked as a scientist at the Army Research Lab, a corporate-style research lab for the US Army. The lab’s goal is to push novel research to higher technology readiness levels, closer to things soldiers might actually use.
Joseph and his colleagues were thinking about electrons, hydrogen atoms, and new 2D materials. The visiting commanders who toured the lab were thinking about something else: future conflict, stronger forces, and better equipment.
Whenever leadership visited, the scientists had to explain why their work should matter to the Army. The relevance of “new 2D materials” was not obvious to those thinking about tactics and battlefields.
That gap created tension, but it also created insight. One day, while explaining how their work might one day change warfare technology, Joseph realized what he really wanted:
Not just to deepen fundamental understanding, but to commercialize science so it actually shows up in products and systems.
That realization nudged him away from the pure research path and toward startups. He started to believe that if you want to push a new technology into the real world in a bold way, a startup is often the most direct path.
If you want to hear him expand on this journey, his story appears in a podcast interview on building Radical AI, where he walks through more of the transition from Army vet to startup founder.
Finding a Mentor Who Demanded Action
Joseph knew he wanted to build a company one day, but he also knew he did not yet know how.
His plan was straightforward: go learn from people who had already built companies and were now backing the next wave. That search led him to New York and to Kevin Ryan, the entrepreneur who built and ran DoubleClick before taking it public, then went on to start Alley Corp, an incubation studio and early-stage investment firm.
In the interview, Joseph told Kevin something bold:
“If you are not investing in materials, you are not going to invest in the future.”
Kevin replied, in effect, “That is a big bet. Come prove it.”
Joseph moved to New York a week later, took a six-month internship at Alley Corp, and treated it as a startup boot camp. Kevin taught him something that now sits at the center of Radical AI’s culture:
Bias to action.
Strategy alone can trap you in meetings and memos. The founders Joseph admires most are the ones who:
- Take action fast
- Watch the real results
- Adjust and move again
As Joseph explains it, successful founders do more than anyone in their space, with more information than anyone else has, guided by a thesis that others have not yet seen.
That mindset is visible in how Radical AI started, how they raised money, and how they run decisions inside the company today. It is also the kind of mindset investors look for, which you can see in writeups like Working Capital Fund’s story on why they invested in Radical AI.
How Radical AI Came Together: Three Co-Founders, One Big Bet
At Alley Corp, Joseph worked alongside another investor, Jorge. While many people were skimming AI headlines, Jorge was reading the research itself. He dug into new architectures, new training methods, and what those might mean beyond chatbots and productivity tools.
One day he turned to Joseph and said:
“AI is so incredible. I am convinced this technology is going to change the world. What I do not understand is why everyone is picking low‑hanging fruit and solving small problems. Why is no one using AI to cure cancer?”
That question stuck.
Joseph’s honest answer was, “I do not know about curing cancer, but that is a really good opportunity.” So the two of them spent the next month and a half reading hundreds of research papers across fields where AI could matter in a deep way.
Materials science kept coming back as a sweet spot:
- The field is fragmented
- Progress is slow and hard to scale
- Timelines are measured in decades, not months
They then went deeper into the intersection of materials science, AI, and robotics. That trail led them to their third co-founder, Herd Cedar, who had built an autonomous scientific lab at Lawrence Berkeley National Lab.
With Herd, the picture sharpened. Together, the three co-founders started to see a future of science that looked very different:
- Less human-only trial and error
- More AI and autonomous systems driving experiments
- A shift from human-driven discovery to AI and autonomy-driven discovery
They all agreed that this future would come whether they built Radical AI or not. That made the decision simple. They wanted to be the ones to build it.
If you are curious about how they describe this self-driving lab approach, there is a good overview in this interview about Radical AI’s vision for materials R&D.
Why Materials Science Is the Bottleneck for So Many Industries
When you zoom out, many of the world’s biggest industries depend on better materials:
- Automotive and aerospace
- Manufacturing and defense
- Climate tech and energy
- Semiconductors and electronics
Underneath these sectors, you often find one big bottleneck: materials R&D. New materials can unlock stronger structures, longer battery life, higher efficiency, or better safety. But again, it usually takes a decade or more to go from first discovery to a scaled system.
Radical AI’s core belief is that materials should not be the blocker. They want AI and autonomy to remove materials as the limiting factor for these industries.
That is why they focus on novel discovery, not just optimization. They are not trying to squeeze 3 percent more performance out of what already exists. They are trying to discover materials that do not yet exist in the world.
From AlphaGo’s “Infamous Move” to Inverse Design
To explain how AI can help with discovery, Joseph often points to AlphaGo, the DeepMind system that beat the world champion in the game of Go.
In one of the most famous games, AlphaGo made an infamous move that shocked commentators. People thought it was a mistake. In reality, the AI had watched so many games and explored so many patterns that it found a move no human had ever considered.
That is the power of indexing at scale.
Science has the same kind of hidden patterns, but the human brain cannot hold millions of papers at once. A scientist is limited by:
- How many papers they can read
- How many simulations they can run and interpret
- How many experiments they can physically test
An AI system is not limited in the same way. It can:
- Read and encode millions of papers
- Run and track billions of simulations
- Help design and interpret thousands of experiments in a feedback loop
This makes something possible that Joseph calls inverse design. Instead of discovering a material first and then going hunting for problems it might solve, you start with the hardest problem, then work backward to design or search for the material that fits.
Human scientists cannot reliably do that today in any reasonable timeframe. AI and autonomy make it far more realistic.
Importantly, Joseph is clear that AI does not replace the scientist. It gives them new tools. It lets them spend more time asking better questions and less time on repetitive lab tasks or literature searches.
Raising a Massive Pre-Seed Round in 45 Minutes
When the three co-founders were ready to leave Alley Corp and turn Radical AI into a standalone startup, they knew they would need real capital. Materials science is expensive. A full-stack solution that blends AI models, robotics, and physical labs is even more expensive.
They built a 100-page pitch deck that covered:
- Why now is the right time for this technology
- Where AI, materials science, and robotics had reached
- Why an interdisciplinary team was needed for the problem
- How their vision could shorten discovery timelines and lift entire industries
They brought it to Kevin Ryan and explained that they planned to raise a large pre-seed round, with Alley Corp as one of several investors.
Kevin’s response was blunt and generous:
“You are not going to raise money anywhere else. I want to give you all of it.”
They closed the entire pre-seed round in about 45 minutes, with Kevin as the only investor. That kind of speed came from years of trust, clear conviction, and a thesis bold enough to justify a big check.
Building a Mission-Driven Startup Culture
Money alone does not build what Radical AI is trying to build. Culture does.
From the start, Joseph and his co-founders decided to filter every hire through a simple question:
“If you are looking for a job, this is not the place for you. You are incredibly intelligent, you can go get employed other places. If you are looking for a mission, then you should come work here.”
Everyone who joins is expected to care deeply about what the company is trying to do, not just about their own role or title. That allows them to push hard, accept uncertainty, and stay through the inevitable setbacks of a frontier startup.
A few core cultural traits stand out:
- Comfort with failure: Trying hard things means some projects will not work. At Radical AI, that is called learning, not personal failure.
- First-principles thinking: Teams are expected to ask “why” often and rebuild their approach when needed.
- Relentless pursuit: They push aggressively on technology and process, knowing that the mission is bigger than any single experiment.
The 51% Rule: How Radical AI Makes Decisions
To keep that bias to action alive, Radical AI borrowed something inspired by SpaceX and adapted it to their world: the 51% rule.
The rule is simple:
When you are at 51 percent confidence on a decision, make the decision.
The alternative is dangerous. If you wait for 100 percent certainty, you will often wait forever. Startups that wait too long get stuck.
At Radical AI, they look at two things to decide whether they are at 51 percent:
- How big is the decision?
Is it about a daily process or a move that could shape the next decade? - What are the risks if we are wrong?
What is the worst realistic outcome, and can we recover?
The team still does research, debates pros and cons, and gathers data. The difference is that once they hit that 51 percent threshold, they pull the trigger.
A few key ideas sit under the 51% rule:
- Many decisions can be reversed or adjusted, so speed matters.
- Thinking for two weeks or two years does not guarantee a correct choice.
- The cost of waiting often outweighs the cost of being wrong and then adapting.
Because they decide this way, failure is baked into how they operate. You will fail at something if you work at Radical AI. That is expected. What matters is how fast you learn and how honestly you update your approach.
Never Stop Collecting Dots
Toward the end of his story, Joseph talks about one of his favorite quotes from Steve Jobs:
“You can never connect the dots looking forward. You can only connect them looking backwards.”
Looking back, his path now feels strangely tight:
- A dad who pushed him to be world-class in something he loves
- Military service that taught him to put mission above ego
- Research at Rice that grounded him in real science
- The Army Research Lab that revealed the gap between lab and impact
- Alley Corp and Kevin Ryan that trained his startup instincts
- Long reading sessions with Jorge that led to Radical AI itself
At the time, none of those steps felt obviously connected. They were just dots. He kept moving, kept asking where he could use his skills in a deeper way, and trusted that the pattern would show up later.
For anyone building a startup today, that is a calming thought. You do not need every answer up front. You just need to keep collecting dots and refuse to give up.
If you want to follow Joseph’s ongoing work and reflections, you can find him on LinkedIn as the CEO of Radical AI.
A Quick Note for Builders: Shipping Still Matters
Radical AI is a story about deep tech and long-term horizons. But even the most ambitious startups need to ship working software.
If you are a developer, you might appreciate how Joseph’s bias to action pairs with a tool like Laravel. The Laravel ecosystem and cloud platform help you build, deploy, and monitor web applications with less friction, so you can spend more energy on the real problem you are solving.
Great ideas do not mean much until they are out in the world.
Also Read: How Two Brothers Turned Breathwork Into an $85K Startup Story (Without Writing Code)
Final Thoughts: Choose Hard Problems, Not Easy Applause
Most AI startups today are playing with small tools and short feedback loops. There is nothing wrong with that, but it leaves something on the table.
Radical AI is a reminder that AI can and should be pointed at the hardest problems we have. It is also proof that you can build a calm, focused, mission-driven company that is comfortable with risk and honest about failure.
If you are thinking about your own startup, you might ask:
- Am I chasing low-hanging fruit, or a problem that, if solved, could shift entire industries?
- Do I want a job, or do I want a mission that could keep me engaged for decades?
- Where can my skills, curiosity, and courage line up with a problem that actually matters?
AI is powerful. Used well, it can help create a world we do not yet think is possible. The real question is whether more of us will choose to use it that way.
0 Comments