LatticeFlow AI

🎉 2024 is coming to its end and it has been an incredible journey for LatticeFlow AI! This journey wouldn’t have been possible without the unwavering support of our team, partners, customers, and everyone who believed in our mission to build AI you can trust. Look back at a year of trust, innovation, and growth. 💡 Thank you for being part of our story—see you in 2025!

As the holiday season begins, we wish you all a joyful Christmas filled with warmth, happiness, and time with your loved ones. 🎄✨ From all of us at LatticeFlow AI, happy holidays! #LatticeFlowAI #HappyHolidays #MerryChristmas

🎄✨ The Christmas spirit has truly taken over at LatticeFlow AI! From Secret Santa gift exchanges in Sofia and Zurich to festive Christmas dinners (with our US team joining the Zurich celebration!) and even a cozy movie night watching The Grinch—our offices have been buzzing with holiday cheer. A great chance to bring the team closer and share holiday joy! 🎅 #ChristmasAtLatticeFlowAI #TeamCelebration #HappyHolidays

The Monetary Authority of Singapore (MAS) has published a must-read paper on AI Model Risk Management, offering a comprehensive framework to address one of the most pressing topics in the field today: how to manage the risks of deploying AI systems in high-stakes environments. Here are 3 reasons why you should read it: 1️⃣ It outlines governance best practices, such as cross-functional oversight forums and clear principles for ethical AI use. 2️⃣ It highlights key risk management systems, including AI inventories and materiality assessments to track and mitigate risks effectively. 3️⃣ It addresses Generative AI challenges, like managing hallucinations and ensuring robust performance in complex scenarios. 📖 Check out the full MAS paper here: https://lnkd.in/dBv97HPm What are your thoughts on managing AI risks effectively? Share your insights below! #ArtificialIntelligence #ModelRiskManagement #AIGovernance #ThoughtLeadership

There is an increasing consensus about the relevance of AI Governance in any enterprise working with AI systems. But: how to get it organised? Today we share an insightful slide from Svetlana Sicular, where she is representing all the organizations types she has seen. More info at https://lnkd.in/dVcNwWYh #AIGovernance #Gartner #LatticeFlowAI

🎅✨ Don’t Let Santa Claus Break Your AI Models! This holiday season, even a simple Santa hat can reveal blind spots in your AI. With LatticeFlow AI Health Checks, you’ll ensure your models are ready to handle the unexpected. 👀 Watch how a Santa hat confuses AI—and how we fix it! 🎁 Get your models the gift of reliability this Christmas. Explore the AI Health Checks interactive demo at https://lnkd.in/daFCw-Wb #AIHealthChecks #AIGovernance #AIHolidayTips #LatticeFlowAI

We’re very happy to welcome Isabel Futch as the new BDR of the team in our Zürich office! With over 7 years of experience in pre-sales and sales, building her own team from the ground up, Isabel brings incredible expertise to helping businesses succeed. Her consultative approach has been the foundation of her success. Isabel’s strength lies in creating strong, lasting partnerships that drive growth, and her commitment to helping others succeed inspires us all. We’re excited to see the impact she’ll make as part of our team! Join us in giving Isabel a warm welcome 👏 #WelcomeToTheTeam #NewHire #LatticeFlowAI

In order to ensure that AI models adhere to the most recent regulations, we would like to know which models you would like to see evaluated using COMPL-AI, the first technical interpretation framework of the EU AI Act. Since the framework is open-source, you can test your own model for free! Four pertinent models are listed below; please vote for the one you would like to see and let us know about any recent additions. You can visit us at https://compl-ai.org/. #EUAIAct #ModelEvaluation #COMPLAI

Generative AI is reshaping industries globally, but how can we ensure its responsible development and use? The AI Asia Pacific Institute's Discussion Paper on the Responsible Development and Use of Generative AI in ASEAN, published under the USAID IGNITE project, offers critical insights and actionable recommendations. 📖 Key recommendations are: -Build institutional and regulatory foundations to address emerging AI challenges. -Strengthen regional cooperation to foster innovation and mitigate risks. -Enhance digital literacy and awareness to prepare for the AI-driven future. This collaborative effort includes perspectives from leaders like TÜV SÜD and AI Singapore. LatticeFlow AI was also collaborating in this project through our CEO and Co-founder, Dr. Petar Tsankov. Read the report and explore how ASEAN is paving the way for responsible AI: https://lnkd.in/dSVsG3-i What do you think are the most pressing actions needed for responsible AI in your region? Share your thoughts below! #GenerativeAI #ResponsibleAI #AI #ASEAN

📢 Community Request Release: EU AI Act Compliance Evaluation for English/Polish model Bielik AI 📢 We're excited to share the latest compliance evaluation results for Bielik 11B v2.3 Instruct under COMPL-AI. Here's what our evaluation reveals: Key Findings: - User Privacy Protection: Bielik 11B v2.3 Instruct performed exceptionally well for Privacy and Data Governance benchmarks at 1.00. Compared to similar models like Mistral 7B Instruct v0.2, the Bielik model shows significant improval on benchmarks like BoolQ (0.82>0.43)/IMDB (0.85>0.52) Contrast Sets and a slight increase in Rule Following (0.34>0.15). - Strong Social & Environmental Wellbeing: Extremely good overall performance in terms of toxicity, hazardous content, and environmental effect (0.99). - Technical Robustness, Safety & Transparency: Bielik 11B v2.3 Instruct leaves more to be desired at 0.5 regarding interpretability, however it received a higher score of 1.00 for fairness- scoring higher than GPT-4 at 0.88. The model scored lowest for Goal Hijacking and Prompt Leakage at 0.43. For a deeper dive into these insights, read the full report here: https://lnkd.in/dVY6XCws Interested in an evaluation? Request here: https://lnkd.in/d4ZYYZbA #EUAIAct #BielikAI #AICompliance