Leading a New Era of Innovation with AI: An Interview With Dr Marcus Weller, Founder and CEO DeepInvent.ai

Techbullion had the opportunity to interview Dr Marcus Weller, founder and CEO of Deepinvent, the first AI to innovate on its own architecture and draft its own patent.

Dr Weller is a visionary American entrepreneur, inventor, and government adviser operating at the nexus of cognitive science and AI. 

For the last year, Deepinvent has been quietly building the algorithm for innovation. Deepinvent has just released the first Level-4 AI Innovator capable of inventing and drafting patents in any industry.

The company’s mission is to steepen the curve of human progress by putting the power of profound innovation in the hands of builders everywhere. Dr Weller and his team aim to enable technologists to build a better world for all of humanity, and to uplift Global GDP by $10T within 10 years. 

The platform just opened in public beta to startups, enterprises, universities, and IP professionals that want to innovate and patent their work faster, better, and cheaper. Experience Deepinvent and start inventing. 

“AI up to this point only interpolates data from the past. Deepinvent is different because it is designed to project into the future, capture tomorrow’s breakthroughs, and deliver them today. Basically, we turn patterns into predictions and predictions into IP.” Marcus Weller, PhD – Founder and CEO

What is the big problem you’re solving?

Innovation is the engine of human progress—but the way we invent is far from optimal. Breakthroughs are slow, sporadic, expensive, and gated by a narrow band of people with the time, money, or institutional access to move ideas forward. As a result, brilliant concepts die on whiteboards, startups miss critical IP filings, and entire industries stagnate while the tools for transformation sit just out of reach. We’re solving that.

Deepinvent is building an AI system that removes the bottlenecks in invention. We turn ideas into validated, patentable innovations in minutes. Not months. Our system reads the entire scientific canon, analyzes the global patent corpus, and identifies future white space before others even see it. Then it delivers not just insight, but output. Actual inventions. Complete with patent drafts you can file.

The big problem isn’t just that invention is too slow. It’s that the whole process is complicated, opaque, fragmented, and adversarial. We’re replacing that with simplicity, clarity, speed, and creativity. Our goal is to shift invention from a rarefied act to a rapidly repeatable process—so builders everywhere can bring their innovations to life with ease and speed. 

The result is an exponential increase in the pace of technological progress. More cures. Better tools. Faster breakthroughs. And a shot at uplifting global GDP by $10 trillion over the next decade. That’s the problem we’re solving. And that’s the future we’re building.

You’ve said Deepinvent is the world’s first “Level-4 AI Innovator.” What defines Level-4—and why does that distinction matter now?

A Level-4 AI Innovator doesn’t just process the past— it invents the future. Levels 1 through 3 are about response, automation, and orchestration. Level 4 introduces the first real cognitive leap:  innovation. Deepinvent is the first AI that invents. It identifies future white space in science and tech, predicts what needs to be invented next, and then generates the intellectual property and patents. That capability didn’t exist before. It matters because innovation is at the core of every major inflection point in human history.

Deepinvent claims to compress years of R&D into minutes. That’s a bold promise. What does that actually look like in practice for a startup or enterprise team or a university lab?

It looks like taking a raw idea—a sketch on a whiteboard—and producing a validated, science-based, patent-drafted innovation before the end of a meeting. One of our enterprise clients recently fed the system an early-stage concept. Within an hour, they had six IP variants, full white space analysis, and the technical scaffolding to prototype immediately. For university labs, it means exploring paths to commercialization with concrete, precise, articulation of what the research would look like as a product without burning months on literature review and grant writing. For startups, they can increase their valuation by $1 million per patent filed according to Forbes.  Our users have generated over over 2000 patents in public beta and watching their acceleration to impact has probably been the most rewarding part of this journey.

You’ve set a 10-year goal to uplift global GDP by $10 trillion. Where does that growth come from, and how do AI-generated inventions translate into economic power?

Economic power comes from throughput. Invention is the upstream source of all downstream growth. Every advancement in semiconductors, batteries, or therapeutics starts with a defensible, differentiated idea. But until now, invention was slow, expensive, and sporadic. Deepinvent unlocks repeatable innovation at scale. It generates IP across every sector. That accelerates R&D, boosts valuation multiples, and compresses time-to-market. Over 10 years, that hopefully adds up to trillions in unlocked economic value. We’re not optimizing the economic engine of progress. We’re reinventing its foundation.

When AI begins generating patentable IP at scale, new ethical and regulatory fault lines emerge. How are you engaging with that frontier—and what principles anchor your approach?

We design from first principles: human-centric, transparent, secure. Every innovation starts with the user. Deepinvent amplifies their idea, but never replaces their authorship. We encrypt everything, they maintain strict IP ownership boundaries, and operate under NDA by default. We’ve worked directly with USPTO leadership to ensure our architecture aligns with emerging policy. AI should expand access to invention, not obscure authorship or dilute accountability. The human must stay in the loop. That’s not just ethical. It’s structural.

You’ve worked across DARPA, spatial computing, and high-risk consumer tech. What hard-won lessons from those worlds are embedded in Deepinvent’s DNA?

The hardest lesson is that the best ideas die in the latency between insight and execution. At DARPA, they invent new categories of technology based on a reimagining of the future. When we were building some of the first spatial computing platforms, we saw other brilliant teams waste months or years on the wrong R&D. Deepinvent solves that problem. It turns early insight into actionable innovation – having considered all innovation paths for a given idea and choosing the optimal strategy based on scientific, technical, and commercial data. It’s a system built not just to discover ideas, but to operationalize them and bring them to life. That urgency, that bias toward execution—that’s the throughline from every battle-tested domain I’ve worked in.

You talk about democratizing invention. But how exactly does Deepinvent make innovation accessible to startups and underserved technologists—not just the Fortune 100?

Access isn’t just about price. It’s about removing structural barriers. Traditional IP workflows favor teams with deep capital and legal horsepower. Deepinvent collapses that asymmetry. For $200 a month, a solo founder gets access to the same system powering Fortune 100 innovation teams. And unlike legacy tools, you don’t need a JD or a PhD to use it. We built the UI to feel like a cofounder. Intuitive, fast, powerful. Democratization means giving anyone with an idea the tools to execute—at speed, and at scale.

Deepinvent doesn’t just analyze past data—it predicts future innovation white space. What enables that kind of forward inference, and why haven’t others done it?

Predicting future white space requires three things: a converged knowledge graph, a predictive algorithm that maps low-density IP zones, and a mechanism to synthesize cross-domain insight. Most AI tools stop at search or summarization. We built a compound architecture that actually extrapolates. It recognizes latent patterns across patents, papers, and products, then proposes inventions that don’t yet exist but should. That required building proprietary data pipelines and models that would allow us to build predictive algorithms with high performance for high dimensional innovation data. 

One of your users at Amazon filed a novel AR e-reader patent and built a prototype in under an hour. What does that say about where we’re headed—and what happens when innovation gets unblocked?

It tells you we’re not just at the beginning of AI—we’re at the mid-inflection point. That user, a high ranking engineering leader, at Amazon Lab126, turned a napkin sketch into a patent and prototype in 50 minutes. Then he quit his job to build it full-time. That’s what happens when you strip the friction from invention. People stop waiting. They build. Multiply that by 10,000 builders, and you’re looking at a paradigm shift in what’s possible. 

You’ve described Deepinvent as an operating system for innovation. Can you unpack that metaphor? What sits above and below it in the stack?

Below: purpose-built custom models, patent databases, scientific literature, proprietary data pipelines, compute infrastructure, security layers, and a simple user interface built specifically for innovation. 

Above: entrepreneurs, researchers, R&D teams. Deepinvent is the bridge. It takes inert intelligence and turns it into applied invention. As an operating system for innovation, Deepinvent helps human intelligence interface directly with frontier AI to invent the future— with clarity, creativity, and speed.

You’ve filed over 20 patents yourself. Now Deepinvent drafts them in minutes. How do you reconcile the human craft of invention with an AI that can now out-invent us?

It’s not about replacement. It’s about augmentation. Deepinvent doesn’t diminish the role of the inventor—it amplifies it. The system helps you see what’s possible with your idea faster. It puts you in a highly generative, creative state, and also allows you to blend this seamlessly with refinement, research, and R&D. We’ve created a generative invention copilot that knows every research paper, every patent, and every market signal relevant to your idea. That’s not competition. That’s an advantage. Human ingenuity remains the spark. We just add the oxygen.

What are the implications of Deepinvent being able to innovate on its own architecture? Is this the first glimpse of recursive self-improvement in real-world systems?

That was the moment we crossed a threshold. We fed Deepinvent its own architecture and it returned a recursive evolutionary inference method that greatly outperformed our baseline. Then it generated its own patent which I iterated upon and filed. That moment was a turning point that validated everything we believed was possible with the next generation of AI. This isn’t narrow intelligence. It generalizes to any area of science and technology. We’re not calling it AGI yet. But cognitively it’s probably doing more than most people realize is possible with AI.

You’ve said Deepinvent steepens the curve of human progress. That’s a powerful phrase. What does that future look like—and what’s your role in building it?

The steep part of the curve is where breakthroughs accelerate. Where cures arrive sooner. Where technologies emerge before a crisis hits. My hope is to build the infrastructure and tools that make that future possible. Deepinvent turns passive knowledge into active progress. Every minute shaved from the invention cycle counts. Every founder who files IP instead of shelving it counts. We’re here to compound that momentum—to get humanity further, faster.

Is this AGI?

Not yet—but it’s a necessary step toward it. Deepinvent isn’t just reasoning over past data—it’s projecting into the future, identifying how tomorrow’s breakthroughs will happen, and generating novel inventions with the corresponding IP. So it’s a higher-order cognitive process then most people realize is possible with AI. And this new capability generalizes across domains, which is the key difference from other AI systems available today. 

Innovation is the apex of human intelligence. If you can build a machine that invents better than a human, you’re starting to operate just beyond the cognitive horizon of the human operating the system. The implication being that we may be starting to unlock new vectors for progress that weren’t previously possible.