GetSeen Ventures

On 25th June, GetSeen Ventures hosted an exciting event exploring the transformative potential of Generative AI in Life Sciences in partnership with Bruntwood SciTech, Kilburn & Strode, and Stevenage Bioscience Catalyst. The day offered invaluable GenAI insights from industry experts and investors, whilst providing a platform for aspiring entrepreneurs and innovative founders to pitch their AI/GenAI startups to a panel of VC investors. The event commenced with an outstanding AI + Life Sciences IP masterclass, delivered by Alexander Korenberg and Sarah Lau from Kilburn & Strode, tailored for aspiring entrepreneurs and early-stage startup founders attended by 25 aspiring entrepreneurs and life sciences founders. Key Highlights from the Fireside Chat: The Fireside chat was moderated by Aline Charpentier (Bruntwood SciTech) with Marco Delise, PhD (Stevenage Bioscience Catalyst), Isabel Zhang (Deeptech Investor, HCVC), Sonal Pai (Digital Biology/Biotech Investor, First Spark Ventures), and Dr Raminderpal Singh (Life Sciences Domain Expert) as panellists. The panel discussed transformative potential of LLMs and the AI adoption in Biotech SMEs to help accelerate research and successful commercialisation in the life sciences sector. Isabel Zhang discussed how deeptech investors evaluate the evolving demographics of AI/GenAI startups and emphasising the importance of large and appropriate datasets for commercial use cases, access to large amounts of compute power and the potential of GenAI in accelerating the life sciences x robotics segment. Sonal Pai suggested looking beyond the AI x drug discovery use cases and explored the potential utilisation of GenAI in accelerating clinical trials management and personalised pharma marketing use cases. Marco Delise, PhD stressed that aspiring entrepreneurs should have a good data strategy to amass large and well governed data sets, robust business and commercial models when planning GenAI ventures in the life sciences sector. All panellists strongly emphasised that founders should aim to have a balanced mix of AI engineers, bioinformaticians, and life scientists trained to drive commercial success in this sector and challenges of talent for early stage startups with domain and generative ai experience. We were also delighted to hear pitches from six outstanding founders in AI/GenAI life sciences, covering areas such as drug discovery, bioprocess development, rapid diagnostics, and the creative (science communication space. Our gratitude extends to Chris Bakal, Matt De Vries, Alex Sheppard, Louis Brookstein, Alex Ramirez, Rory Ryan, and Ashkan Dabbagh. A heartfelt thank you to our panellists, startup founders, aspiring entrepreneurs, angel investors, and our event partners: Bruntwood SciTech, Kilburn & Strode, and Stevenage Bioscience Catalyst for making this event a resounding success! HCVC, First Spark Ventures #GenAI #London #Startups #LifeSciences #GetSeenVentures #Innovation

GetSeen Ventures continues to deliver exceptional value to our GenAI Bootcamp 2024 participants! We're thrilled to announce our keynote speaker for the GenAI bootcamp: Ahmed M.Raafat, Principal Solutions Architect @ Amazon Web Services (AWS) Topic: "How to productionise Your Generative AI Application with Amazon Web Services (AWS)" 📅 18th September | ⏰ 1300 BST This high-caliber session exemplifies our commitment to connecting you with industry leaders and cutting-edge insights. Ahmed's expertise in Amazon Web Services (AWS) and generative AI applications promises to be invaluable for your startup journey. To all the bootcamp participants: Don't forget to mark your calendars and come prepared with your questions!

GetSeen Ventures is launching GenAI Research Paper Club! GetSeen Ventures AI Research paper club will bring together GenAI technologists, researchers, and founders with a deep passion for GenAI to discuss and analyse cutting-edge and fast-evolving GenAI research across the globe. This exclusive group will meet once a month (virtual) to dive deep into groundbreaking research papers, sparking discussions and uncovering fresh insights. If you're looking to challenge your thinking and stay ahead of the GenAI curve, this group is for you! The sessions will comprise of one topic every month - presentations by GenAI Researchers - breaking down the papers conceptually, followed by a peer discussion Register your interest at this link: https://bit.ly/3MMrbw7

This week at the GetSeen Ventures GenAI bootcamp our guest speaker was Vinitra Swamy, a machine learning PhD researcher at EPFL. Vinitra's research on eXplainable AI and ML for education and in the webinar she spoke about the future of human-centric eXplainable Artificial Intelligence (XAI). Top 3 takeaways from this webinar: Clarity is key in AI for education: We explored how AI explanations need to be consistent, easy to understand, and actionable for educators. The focus on 'Explainability Needs' highlighted the importance of accuracy and real-time feedback in educational AI tools. Explainable AI (XAI) builds trust: In sensitive fields like education, XAI is crucial for creating trustworthy AI systems. Transparency in AI decision-making processes allows for confident implementation and continuous improvement. Navigating explainability techniques: We delved into the challenges of different explainability methods like SHAP and DiCE. Their occasional disagreements underscore the need for reliable explainability approaches, especially in high-stakes fields like healthcare and education. These insights reinforce the importance of human-centric AI development. As we continue to integrate AI into education, keeping these principles in mind will be crucial for creating effective, trustworthy systems that truly benefit students and educators. #GenAI #GetSeenVentures #GenAIbootcamp 

GetSeen Ventures GenAI bootcamp Week 4 For Week 4 of the GenAI bootcamp, GetSeen Ventures recently had the pleasure of hosting Dan Figueiredo, MBA, the Global Technology Strategist at Microsoft enlightening the participants about the current challenges and opportunities using AI. Key takeaways from Dan’s webinar: Bridging the Hype Gap: Generative AI is advancing rapidly, but there is a significant gap between the hype and actual outcomes. Without careful management of expectations, this gap can lead to market disappointment. Dan emphasised the need for a critical assessment of AI’s return on investment, focusing on tangible benefits rather than just potential. Credibility and success in this field depend on a grounded approach. The Power of Small Language Models: While Large Language Models dominate AI, Dan highlighted the growing potential of Small Language Models (SLMs). These specialised models often outperform larger ones in specific tasks, offering sustainable solutions by reducing training, maintenance costs, and energy consumption. This challenges the notion “Bigger is better" mindset and suggests an alternative path for AI development. The Need for Ethics in AI Development: As AI technology advances, the call for responsible development becomes increasingly urgent. Dan argued that AI should prioritise societal needs over commercial interests. Governments and regulatory bodies must ensure that AI development adheres to ethical standards, reinforcing our duty to build systems that benefit society. Thanks a lot to Dan Figueiredo, MBA for this insightful session! #GenAI #GetSeenVentures #GenAIbootcamp 

Great to see our co-founder Tanmay Gupta, PhD exploring the Austrian and wider European AI/DeepTech ecosystems in the IECT - Hermann Hauser Summer School programme!

GetSeen Ventures Hosts Groundbreaking AI Session: Unveiling the Future of LLMs For Week 3 of the GenAI bootcamp, GetSeen Ventures recently had the pleasure of hosting Luke Johnston, a distinguished alumnus of University of Oxford and University of Cambridge, for a session on "Building with LLMs". The event offered our attendees a glimpse into the revolutionary advancements shaping the future of large language models (LLMs). We're excited to share three key takeaways that underscore the rapid evolution of AI development: Advanced Techniques in AI Training: LoRA, QLoRA, and RetNets Luke shed light on innovative methodologies aimed at enhancing the efficiency and performance of LLMs. Low Rank Adaptation (LoRA) and its quantised counterpart, QLoRA, introduce a novel approach to fine-tuning, significantly reducing trainable parameters and memory footprint. Additionally, Retentive Networks (RetNets) offer a promising alternative to traditional attention mechanisms, potentially revolutionising sequence modelling with improved efficiency. Enhancing AI Performance: Direct Preference Optimisation and Retrieval Augmented Generation The session explored techniques for aligning AI more closely with human preferences and expanding its knowledge base. Direct Preference Optimisation (DPO) streamlines the process of incorporating human feedback, whilst Retrieval Augmented Generation (RAG) enables LLMs to access external knowledge during inference, enhancing accuracy and relevance in specialised domains. Explainable AI and Multimodal AI: Towards Greater Transparency and Versatility As AI systems become increasingly integrated into sensitive areas, we learned about the critical importance of Explainable AI (xAI). Techniques such as attention visualisation and feature importance analysis are crucial for fostering trust and accountability. Moreover, the advent of Multimodal AI promises more versatile systems capable of processing and generating multiple data types, marking a significant step towards more human-like AI interactions. At GetSeen Ventures, we believe these advancements not only push the boundaries of AI capabilities but also address critical aspects of efficiency, transparency, and real-world applicability. We're thrilled to facilitate discussions that keep our community at the forefront of technological innovation. #GenAI #GenAIbootcamp #GetSeenVentures 

Unveiling the Future of AI: Insights from Our Latest Webinar on Retrieval-Augmented Generation For week 2 of the GenAI bootcamp, GetSeen Ventures hosted an enlightening webinar featuring Yuan He from the Department of Computer Science at the University of Oxford. The session, titled "Retrieval-Augmented Generation (RAG) for LLMs” offered invaluable insights for professionals in AI and natural language processing. Our attendees working on generative AI solutions found the session particularly illuminating. The discussion on mitigating hallucination in AI, where models generate plausible yet false information was especially impactful. We're excited to share three key takeaways that resonated strongly with our followers: 1.⁠ ⁠The Power of Sparse Retrieval and Inverted Indexing Yuan introduced the concept of Sparse Retrieval, which utilises an Inverted Index for external document indexing. This method, which maps specific keywords to the documents they appear in, offers an efficient alternative for scenarios with a narrow keyword range. 2.⁠ ⁠Enhancing Retrieval with Pre- and Post-Query Processing The webinar highlighted the often-overlooked importance of pre- and post-query processing in improving retrieval effectiveness and model output quality. Techniques such as query routing, rewriting, and expansion, along with post-processing methods like reranking and summarisation can significantly enhance the accuracy of AI models. 3.⁠ ⁠GraphRAG: A Cutting-Edge Approach to RAG Perhaps the most exciting development discussed was GraphRAG, which incorporates knowledge graphs into the retrieval process. This method offers deeper contextual understanding and more precise information retrieval, particularly beneficial for specialised fields like medical research or niche technical domains. We extend our heartfelt thanks to Yuan He for sharing his expertise and to all who attended this enlightening session. #GenAI #GenAIbootcamp #GetSeenVentures

Last week, GetSeen Ventures launched its inaugural GenAI bootcamp, welcoming student developers from top UK universities. The program aims to empower UK students with the skills to build real-world GenAI applications. The bootcamp kicked off with an insightful session on "Factuality and Plausibility with Language Models" led by Moy Yuan, a PhD researcher at the University of Cambridge. Yuan began with an introduction to Natural Language Processing (NLP) and Large Language Models (LLMs) and then delved into the crucial concepts of factuality and plausibility, highlighting their importance in developing reliable and trustworthy AI systems. Key insights from Moy's talk included: - Factuality vs. Plausibility: Models excelling in factual accuracy don't always perform well in plausibility. AI practitioners face the challenge of balancing these aspects or using specialised models for each. - RAG models' limitations: While popular for integrating external knowledge, Retrieval-Augmented Generation models often miss critical information, especially with complex inputs. This can lead to plausible but potentially inaccurate outputs. - Practical tip: Adopt an agentic workflow approach for efficient LLM use, even without high-end GPUs. This involves using LLMs as collaborative agents, breaking tasks into manageable components, and leveraging different models' strengths. - Future of NLP & LLMs: Balancing factuality and plausibility remains a key challenge. Ongoing innovation is crucial to address these complexities and create more reliable AI systems. We would like to extend our gratitude to Moy Yuan for taking out time from his busy schedule and give this amazing webinar. These insights are crucial for developing reliable, trustworthy AI systems. Excited to see how our student developers will apply this knowledge! #GenAI #GetSeenVentures #GenAIbootcamp 

Great to see the GetSeen Ventures founders and entrepreneur community event was a grand success in Frankfurt! Looking forward to new collaborations and exciting events ahead :) Thanks to all the organisers and attendees! Daniel Gerald Orbegoso Barrantes Daniel Pietschmann Hamed Roknizadeh Huma Erinc Ercan Joshua Friedrich Oliver Friedrich Timo Müller-Tribbensee Tobias Stein Maximilian Päzolt 💡💻Marcel Isbert Fayssal El Mofatiche, CAIA Prachi Jain #AI #GenAI #getseenventures #Frankfurt #Germany #Startups