A bottle opener hidden inside a flip-flop was Steve Hafif’s first encounter with the world of patents. While working summers at his father Sam’s accessories manufacturing business, Hafif was given an unusual assignment: identify potential licensing partners for a patented product that blended novelty with practical design. At the time, Hafif was still in high school, learning firsthand how intellectual property functions not only as a legal construct, but also as a signal of innovation and commercial intent. The licensing effort never materialized, but the experience left a lasting impression.
That early exposure resurfaced years later during Hafif’s studies in data science at Drexel University.
Patent insider
Who: Steve Hafif
Innovation cred: Founder, CEO of Cypris, self-professed flop at licensing father’s flip-flop/can opener patent during high-school summer job
Favorite invention: Patent 821,393 for the Wright brothers’ “flying machine.” Orville and Wilbur filed their application on March 23, 1903. The patent was granted May 22, 1906, more than three years later. According to the patent description: “Our invention relates the class of flying machines in which the weight is sustained by the reactions resulting when one or more aeroplanes are moved through the air edgewise at a small angle of incidence.”
Through the school’s cooperative education program, he worked with Exelon’s R&D innovation team, collaborating with engineers on internal research initiatives focused on protecting, analyzing, and commercializing new technologies. Patent data became central to this work, both as a technical input and as a way to understand broader innovation landscapes.
That intersection of data science, R&D, and intellectual property ultimately led to the founding of Cypris in 2020. Today, Hafif serves as Founder and CEO of the company, which helps corporate R&D teams extract actionable intelligence from patent data alongside scientific literature, market signals, and other innovation sources. Cypris is designed to support researchers as they evaluate emerging technologies, identify strategic opportunities, and bring new inventions to market. Over the past three years, the company has more than doubled its customer base annually.
The path to that growth, however, was not linear. Cypris initially launched as a marketplace intended to license university-developed IP to corporate R&D teams. While demand for direct licensing proved limited, those early conversations revealed a deeper opportunity: R&D leaders were just as interested in the signals embedded within patent data as the patents themselves. That insight prompted a pivotal shift from IP marketplace to intelligence platform, reshaping Cypris’ mission and trajectory.
IFI CLAIMS spoke with Hafif in December to discuss the evolution of Cypris, the lessons learned from its early pivots, the role of patents as a foundational data asset for AI-driven research, and why Cypris built its platform on IFI’s patent data. The interview has been edited for clarity and length.
IFI CLAIMS: Tell us about yourself and how you became involved in the patent industry.
Hafif: My first exposure to patents came while I was still in high school. One summer, I was working at my father’s accessories manufacturing company. He held a patent for an unusual product, a flip-flop with a retractable bottle opener embedded in the sole, but it was not being actively commercialized. He asked me to explore licensing opportunities for the IP.
That experience required me to engage directly with patent documents for the first time. I read the patent, learned how to interpret the claims, and approached brands like Oakley and Billabong to explore licensing discussions. While none of those efforts ultimately came together, the process gave me an early, practical understanding of how patents function as both legal instruments and commercial signals.
Later, while studying data science at Drexel University, I participated in a cooperative education program that placed me on the R&D innovation team at Exelon. There, I worked closely with engineers on internal research projects aimed at developing and commercializing new IP. Given my earlier exposure to patent data, I gravitated toward patent datasets, helping to build internal search and discovery tools and spending significant time analyzing how that data could inform technical and commercial decision-making. That combination of hands-on R&D experience and deep interaction with patent data ultimately laid the foundation for Cypris.
IFI CLAIMS: So you went from college straight into founding your own business? What was that like?
Hafif: Entrepreneurship was always the end goal for me. I come from a family where building businesses is common, and from an early age I knew I wanted to create something of my own. I did not initially know what the company would be or which market it would serve, but I was drawn to the process of building something durable and meaningful.
Technology, and particularly AI, stood out as an area where that ambition aligned with long-term momentum. Even early on, it was clear that AI would fundamentally reshape how work gets done across industries. During college, I spent time thinking about those shifts and reading authors who explored them in depth. That body of work helped crystallize the idea that AI would increasingly converge with real-world decision-making in complex domains.
Cypris emerged from that intersection: a desire to build something meaningful, a deep interest in AI, and a conviction that intelligence infrastructure, not just tools, would define the next generation of enterprise software.
The goal was to eliminate the need for R&D teams to stitch together insights across disconnected platforms. Instead, Cypris provides a single, consolidated intelligence layer where those questions can be explored in full context.
IFI CLAIMS: What is your founder’s story? And what exactly does Cypris do?
Hafif: Cypris initially started as an IP commercialization marketplace. In 2020, I spent much of the year working with university technology transfer offices at institutions such as Penn, Harvard, and MIT to list patents that were available for licensing. The idea was to aggregate high-quality academic IP and connect it with corporate R&D teams that might want to license and commercialize it.
What quickly became apparent was a mismatch between supply and demand. Corporate R&D teams were largely uninterested in licensing university patents directly, viewing many of them as too early stage or disconnected from near-term commercial needs.
However, those conversations revealed something more interesting. While the patents themselves were not compelling as licensing assets, the signals embedded in the data were. R&D leaders were surprised to see where top universities were investing their research efforts and how those investments mapped onto emerging technology landscapes. That insight prompted a pivot.
We shifted from a licensing marketplace to a market and technology intelligence platform, using patent data as a foundational signal to help R&D teams understand where innovation was happening, who was driving it, and how it connected to their strategic priorities. Early customers, including teams at Hyundai and Catalent, validated this approach. Over time, Cypris evolved into an intelligence platform purpose-built for corporate R&D.
IFI CLAIMS: What problem do you solve for your clients?
Hafif: The challenges vary, but the underlying pattern is consistent. R&D teams are constantly trying to answer complex questions that may be technical, commercial, or strategic in nature. The information required to answer those questions is typically fragmented across many tools and data sources. Over time, we expanded beyond patent data into scientific literature, startup and commercial datasets, market signals, and consumer sentiment. The goal was to eliminate the need for R&D teams to stitch together insights across disconnected platforms. Instead, Cypris provides a single, consolidated intelligence layer where those questions can be explored in full context.
IFI CLAIMS: What are the specific differentiators?
Hafif: Our primary differentiator is focus. From early on, we made a deliberate decision to build exclusively for corporate R&D teams. By narrowing the customer profile, every aspect of the platform, from the interface to the ontology to the way data is structured, could be designed around how R&D teams actually work.
Most incumbent solutions were built for IP and legal professionals. They are powerful, but often complex and unintuitive for non-IP users. We took the opposite approach by making sophisticated intelligence accessible without requiring patent expertise.
IFI CLAIMS: Is there anything else important to know about Cypris?
Hafif: Cypris is an AI-first company by design. From the outset, we built the platform around advanced AI capabilities rather than layering them on later. We work closely with OpenAI, Anthropic, and Google, and we integrate new models into the platform as soon as they become available.
This matters because AI is not static. Models improve rapidly, and the ability to adopt them quickly and intelligently directly impacts product quality. Many legacy providers are still retrofitting AI into systems that were not designed for it, which creates friction. Our architecture allows us to move at the pace the technology itself is moving.
From early on, we made a deliberate decision to build exclusively for corporate R&D teams. By narrowing the customer profile, every aspect of the platform, from the interface to the ontology to the way data is structured, could be designed around how R&D teams actually work.
IFI CLAIMS: Why do you work with all three AI models?
Hafif: At this scale of investment, it is unrealistic to assume there will be a single long-term winner. Each model provider excels in different areas. Rather than betting on one, we designed Cypris to dynamically leverage the best model for a given task.
For example, we have found that some models perform exceptionally well with technical literature, while others are particularly strong in market analysis. When a user asks a question, Cypris can draw on multiple models and synthesize their strengths into a single response. This approach is technically demanding, but it ensures that users always benefit from the best available AI.
IFI CLAIMS: Any general lessons for other founders?
Hafif: One of the most important lessons came during our early pivot. At the time, I believed we were making strong progress. We had high-quality data and a polished presentation. Then a venture capitalist challenged me very directly. He argued that what we had was vanity, not value.
He framed value very simply: you either make people money, save them money, or save them time. If you are not doing one of those three things, you are not creating real value. That perspective was uncomfortable, but clarifying. It forced me to rethink who we were serving and where we could deliver measurable impact.
IFI CLAIMS: What trends are you seeing in the patent space?
Hafif: I view patents primarily as a data asset rather than a legal construct. From that perspective, they are extremely rich, particularly for AI systems that can extract patterns and signals at scale.
At many large organizations, engineers are rewarded for filing patents rather than commercializing outcomes. That incentive structure can lead to a growing volume of filings that are only loosely connected to real-world value. Distinguishing signal from noise becomes increasingly important.
I expect those incentives to evolve, particularly as AI accelerates product cycles. Over time, fewer but higher-quality patents, supported by clearer evidence of use, would significantly increase the value of the dataset as a whole.
IFI CLAIMS: Why did Cypris choose IFI data?
Hafif: We initially explored building directly from patent office datasets, but it quickly became clear that approach was neither scalable nor sustainable. Maintaining global coverage at the level of quality we needed would have consumed disproportionate resources.
IFI consistently came up through referrals, and once we evaluated the data, the difference was immediately apparent. IFI operates at a level of completeness and reliability that allows us to focus on intelligence, synthesis, and application rather than rebuilding foundational infrastructure. That partnership enables Cypris to treat patent data as one powerful component of a broader intelligence stack.
