We divide these changes into two main categories: Evolution and Revolution. Evolution refers to the continuous improvement of existing processes through AI, enhancing efficiency and effectiveness. Revolution, on the other hand, encompasses groundbreaking transformations that create entirely new opportunities and capabilities. This thesis aims to explore how AI can be strategically applied to drive both evolutionary and revolutionary advancements, benefiting businesses and society as a whole. By supporting founders who embrace AI, we are committed to fostering innovation and pushing the boundaries of what is possible.

Evolution

Augmentation of the Modern Workforce: Workflow Co-Pilots

The current development of AI brings new challenges and opportunities for various industries, primarily by transforming workflows and information retrieval. Evolution refers to the continuous improvement of existing processes, where every workflow evolves as mundane or repetitive tasks are automated. In fields like accounting, insurance claims, or healthcare, AI can anticipate next steps, which previously required time-consuming work and learned intuition. This automation not only saves time but also reduces error rates, allowing employees to focus on more complex and value-added tasks.

AI agents have shown incredible capabilities to automate repetitive tasks. An example of this evolutionary improvement is the use of chatbots and voice assistants that facilitate everyday tasks such as compliance checks or ordering parts in manufacturing. Instead of navigating through multiple systems and making manual entries, employees can make their requests in natural language, and AI takes care of the rest. This not only improves efficiency but also enhances the system's user-friendliness.

The real value of AI lies in its ability to provide information at the right moment. By automating repetitive tasks and delivering timely information, AI helps streamline workflows and enables employees to focus on higher-level strategic tasks, thereby augmenting the modern workforce and driving productivity.

Making Everyone Smarter: Information Retrieval & Enterprise Search

AI's capability to quickly and accurately find and analyze large amounts of information, known as information retrieval (IR), is transforming how we access the right information. This transformation is not just about retrieving data; it's about finding the most relevant and valuable information efficiently. We see a surge of Retrieval-Augmented Generation (RAG) companies in the market, but RAG alone is not a definitive solution. The true value lies in the smartness of how IR is conducted.

At Cherry Ventures, we believe that the quality of IR is paramount. Techniques such as Low-Rank Adaptation (LoRA), embedding information in vector spaces, and advanced prompting methods are crucial in enhancing the effectiveness of IR. While these approaches can be hard to measure, they fundamentally improve the quality of the information retrieved. This is the competitive arena where platforms like You.com, Bing, and OpenAI, as well as specialized industry-specific search solutions, are vying for supremacy.

There are various types of IR: company-specific, industry-specific, and horizontal. It remains to be seen whether a one-size-fits-all solution will emerge or if specialized approaches will dominate. For instance, in the medical field, AI can assist doctors by retrieving relevant patient data and medical research, enhancing diagnostic accuracy and treatment plans. Similarly, in the legal domain, AI helps lawyers quickly find pertinent legal texts and precedents, streamlining case preparation and research. Examples of effective IR approaches include using advanced vector embeddings to map complex medical data, allowing healthcare professionals to access the most relevant patient information and research findings. In the legal sector, sophisticated prompting techniques enable AI to provide lawyers with precise and contextually relevant legal documents and case law.

In summary, while the market is flooded with RAG solutions, the real game-changer is the quality of IR. By leveraging advanced techniques and focusing on the smartness of information retrieval, we can significantly enhance the capabilities of AI in various fields, making everyone smarter and more efficient.

The New Tooling Wave: Taming AI

2024 marks the beginning of a new tooling wave. This wave encompasses the creation of tools to maintain, manage, and set up AI systems effectively, forming what is known as the AI stack. Companies like Hugging Face and Civic AI are leading the charge in providing solutions for quality control, training, and security in AI applications, ensuring that AI models produce reliable and safe outputs, particularly in enterprise environments where quality and privacy are crucial.

A key component of this tooling wave is Machine Learning Operations (ML Ops), which focuses on the end-to-end lifecycle management of machine learning models. This includes automating deployment, monitoring performance, and ensuring continuous updates with the latest data. Another critical aspect is prompt engineering, which involves designing and refining the prompts that guide AI models, thus establishing robust guardrails to ensure accurate and contextually appropriate responses.

These tools permeate the entire AI stack, enabling both revolutionary and evolutionary advancements. For instance, improving techniques like Low-Rank Adaptation (LoRA) to make them faster and more efficient is crucial. Deciding whether to consistently use the same model for embedding in vectorial space or exploring better alternatives can significantly impact performance. Selecting the optimal vector database for specific use cases and optimizing prompt engineering are essential tasks that these tools help manage. Additionally, measuring the effectiveness of these components collectively is vital for achieving optimal results.

The iterative process of discussing and refining prompts is essential, allowing teams to collaboratively improve AI models. Securing data with robust encryption and access control, maintaining data quality through continuous monitoring, and managing complex workflows that integrate multiple AI models are all vital components of this new tooling wave. By leveraging these tools and practices, organizations can optimize the performance and efficiency of their AI applications, driving innovation and achieving strategic objectives.

Revolution

The Interface Leap: Your Personalized Access to Information

Revolution describes groundbreaking changes that open up entirely new possibilities. A key aspect here is the interface leap, where AI significantly transforms how we access and interact with information. This shift seems to challenge 60 years of user experience (UX) knowledge by bringing us back to text-based interfaces. What comes next—voice interactions, a blend of clicks and text, or something entirely new?

AI systems now provide personalized, concise summaries and relevant information swiftly, eliminating the need to search through databases manually. This leap is evident in fields like customer care and legal services, where AI can aggregate and present information efficiently, saving time and improving accuracy. Soon we will use our voice regularly to complete tasks, making work on the go more seamless.

To support this evolution, we need tools that can seamlessly handle these varied interactions, ensuring that AI systems remain efficient, accurate, and user-friendly. They will democratize access to information but also enhance decision-making processes by delivering high-quality, relevant data instantly through multi-modality. As we move forward, it will be crucial to develop and refine these tools, balancing the rich history of UX design with the innovative possibilities AI brings. This ongoing evolution will shape how we interact with technology, making information more accessible and interactions more natural and effective.

An Age of Supercreation: Supercharging Innovation & Creativity

AI ushers in an age of supercreation, where it supports and enhances human creativity and innovation without replacing the unique contributions of artists, UX designers, or founders. While AI and LLMs act on past data, their role is to assist in the ideation process by providing valuable insights and suggestions. This dynamic is akin to the relationship between a chess player and a computer—the best outcomes arise from human evaluation of the strategies suggested by AI.

By automating routine tasks and offering insightful data analysis, AI frees individuals and companies to focus on innovative and creative endeavors. This includes developing new products, improving services, and finding novel solutions to complex problems. AI's ability to identify patterns and opportunities that may not be immediately apparent helps drive innovation forward at an unprecedented pace.

The interaction between humans and AI is crucial. AI tools are integrated into workflows, providing support where it matters most—whether in brainstorming sessions, design processes, or strategic planning meetings. This collaborative dynamic allows humans to leverage AI’s strengths, such as data processing and predictive analytics while applying their own creativity and critical thinking.

In essence, AI supercharges the ideation process, enhancing human capabilities rather than replacing them. This synergy between human ingenuity and AI’s capabilities defines the new era of supercreation, ensuring that innovation is both supported and accelerated by advanced technological tools.

The Comeback of Hardware: Making Metal Smart

The revolution in AI is bringing a comeback in hardware innovation. AI is not just about software; it also allows hardware to do things it was not able to do before. Integrating AI into various devices and machinery, such as drones, wearables, consumer devices, and medical devices, makes them more efficient and capable. Smart hardware can adapt to different tasks, learn from interactions, and improve performance over time. This transformation is particularly significant in industries like manufacturing, healthcare, and transportation, where smart hardware can lead to more efficient operations, better patient outcomes, and safer, more reliable transportation systems.

Cherry firmly believes the advancement of AI brings both evolutionary and revolutionary changes. Through the automation and improvement of existing processes, as well as the ability to efficiently process large amounts of information, AI opens up new possibilities and challenges. Companies and societies must actively engage with these changes to fully leverage the benefits of AI and address the associated challenges.

Tune into the newest episode of The Edge to learn all about our AI investment theses with Jasper Masemann, investment partner, and Lutz Finger, venture partner at Cherry Ventures!