Syntro

We are very excited that Syntro has been selected to be a part of the Female Founders #GrowF program! Dive into the article 👉 https://lnkd.in/gTuexK4E 👈 to learn more about Syntro and meet the co-founders, Agata Staszak & Wojtek Śmigiel on their mission to enhance the efficiency and reproducibility of life science research.

We're thankful for the opportunity to connect with inspiring individuals and discuss the potential of task-oriented AI agents for automating knowledge management in research. Looking forward to future collaborations and events!

Exciting news for genomics research!🧬 With GeneGPT, researchers can now access National Center for Biotechnology Information (NCBI) database insights with natural language queries. Thanks to the efforts of Qiao Jin, Yifan Yang, Qingyu Chen, and Zhiyong Lu, PhD FACMI, navigating genomics data has never been easier or faster. Here's how it works: 1️⃣ GeneGPT incorporates a specially designed prompt, comprising documentation and practical demonstrations of NCBI API usage. This equips the model with the necessary guidance to navigate and extract relevant information from the NCBI database. 2️⃣ GeneGPT integrates an inference algorithm into the decoding process of the LLM (in this case Codex). This algorithm is responsible for detecting and executing API calls to access relevant information. GeneGPT has been put to the test on GeneTuring, a QA benchmark for genomics, which consists of 600 questions covering a spectrum of genomics topics. These questions are categorized into four modules, including gene nomenclature, genomics location, functional analysis, and sequence alignment, addressing twelve common tasks encountered in genomics research. GeneGPT outperforms existing models by a significant margin, achieving an average score of 0.83 on eight tasks within the GeneTuring benchmark 🏆, leaving behind other retrieval-augmented LLMs and biomedical LLMs, including Bing (0.44), BioMedLM (0.08), BioGPT (0.04), GPT-3 (0.16), and ChatGPT (0.12). Curious to see GeneGPT in action? Check it out here: https://bit.ly/3IGPbyE Or explore the details in the paper: https://bit.ly/3PqrtKR #GeneGPT #LLMs #Genomics #AI #NCBI #GeneTuring

Exciting news for genomics research!🧬 With GeneGPT, researchers can now access National Center for Biotechnology Information (NCBI) database insights with natural language queries. Thanks to the efforts of Qiao Jin, Yifan Yang, Qingyu Chen, and Zhiyong Lu, PhD FACMI, navigating genomics data has never been easier or faster. Here's how it works: 1️⃣ GeneGPT incorporates a specially designed prompt, comprising documentation and practical demonstrations of NCBI API usage. This equips the model with the necessary guidance to navigate and extract relevant information from the NCBI database. 2️⃣ GeneGPT integrates an inference algorithm into the decoding process of the LLM (in this case Codex). This algorithm is responsible for detecting and executing API calls to access relevant information. GeneGPT has been put to the test on GeneTuring, a QA benchmark for genomics, which consists of 600 questions covering a spectrum of genomics topics. These questions are categorized into four modules, including gene nomenclature, genomics location, functional analysis, and sequence alignment, addressing twelve common tasks encountered in genomics research. GeneGPT outperforms existing models by a significant margin, achieving an average score of 0.83 on eight tasks within the GeneTuring benchmark 🏆, leaving behind other retrieval-augmented LLMs and biomedical LLMs, including Bing (0.44), BioMedLM (0.08), BioGPT (0.04), GPT-3 (0.16), and ChatGPT (0.12). Curious to see GeneGPT in action? Check it out here: https://bit.ly/3IGPbyE Or explore the details in the paper: https://bit.ly/3PqrtKR #GeneGPT #LLMs #Genomics #AI #NCBI #GeneTuring

Have you ever wondered about the source behind AI-generated content or tested generative models and caught them hallucinating? Large Language Models (LLMs), while powerful, often leave us questioning the origins of their insights. That's where Retrieval-Augmented Generation (RAG) steps in! By incorporating RAG into your AI systems, you can effectively 'train' LLMs to draw information from specific, reliable sources, enhancing their credibility, usefulness, and traceability of sources. The RAG workflow consists of 3 parts: 1️⃣ Retrieve: The user query is utilized to retrieve relevant context from knowledge sources or a database. 2️⃣ Augment: The user query and the retrieved context are inserted into a prompt template instructing the LLM. 3️⃣ Generate: Finally, when the prompt is fed to the LLM, a response based on the provided context is generated. Using this method of retrieving information from your database, you'll achieve quick access to vast amounts of relevant information with traces to real information. 💡 At Syntro, we specialize in utilizing LLMs to enhance lab workflows. If you're interested in learning how to incorporate RAG into your research, please feel free to schedule a meeting with us: https://bit.ly/syntro. #Biopharma #Biotech #AI #LLMs #RAG

Thinking of using LLMs in your lab? 🤔 According to Deloitte, over 90% of respondents from the biopharma and medtech industries expect generative AI to significantly impact their organizations in 2024. Decision makers need to develop new strategies to leverage the rapid advancements in AI and Large Language Models, often with little or no prior experience in generative AI. When used effectively, LLMs can become a productivity powerhouse. Their applications range from simple text-based operations to complex task execution and reasoning. ➡️ At Syntro, we specialize in utilizing LLMs to enhance lab workflows. If you're interested in exploring LLM solutions tailored to your research, please feel free to schedule a meeting with us: https://bit.ly/syntro. #Biopharma #Biotech #AI #LLMs

🚀 Exciting times at Syntro. We just launched a pilot with the brilliant minds at Genomtec SA! We are on our way back from Wrocław, where we met the team at the cutting edge of clinical diagnostics. During our visit, we led a workshop on how LLMs can revolutionize research processes. As we move forward, the team will be testing our product and helping us shape the future of AI-assisted research. 🔍🔍 We are grateful for the warm welcome! #AI #LLMs #ClinicalDiagnostics #Genomtec #Pilot

Syntro takes home 2nd place! Congrats to everyone in the #eitdigital category, and everybody at the #jumpstarter grand final! The pitches were excellent and the competition was fierce, but nothing could beat the #teamplay you could feel in the air. It's great to be a part of such a welcoming community!

You can read about Syntro's cofounding story in the November edition of MyCompany Polska!

We are super excited for Athens next week! The #eitdigital roster is so cool and diverse! Can't wait to see you all in person!