Hi PH! Niya here - one of the people working on Seia The problem 🧩 🔬 AI models tend to work as intended in lab environments. The problems start emerging only when you deploy models into the real world. This is when they start creating tons of harmful outcomes because of the way they learn from their context. ⚖️ New regulations like the European Union’s AI Act will require companies to be able to manage emerging risks and ensure human oversight. 📈 There are multiple monitoring and observability tools out there, but they are all focusing on the model performance and accuracy. 🖼️ Trying to understand your deployed model’s impact and looking only at its performance and accuracy is the same as looking at a 2-storey high painting through a pinhole. No matter how hard you try, you’ll never get the full picture. Solution 💡 For the past few months we’ve been working towards a solution that enables organisations to get the full picture by not only understanding the performance of their models, but also how it impacts the business and involved humans. Seia offers a comprehensive solution: 🔍 Tracks deployed models' performance, ensuring you have a clear understanding of their behaviour in real-world scenarios. 🔒 Collects and analyses human feedback to enhance outcome accuracy and promote fairness by combining mathematical fairness metrics with contextual insights from users. 💸 Adds a layer of systemic risk management by providing an overview of models' impact on critical KPIs. Proof of concept 🦾 To showcase the power of Seia, we’re initially targeting model and business performance overview and ground truth delay. Your Feedback Matters 🦻 We're at an early stage of development, and your valuable feedback is instrumental in refining Seia. We invite you to be a part of our journey in building a solution that ensures deployed AI systems perform as intended across any context and user interaction. Join our community and sign up for free access to Seia's proof of concept here: https://seia-reliableai.deus.ai/ Let's unlock the true potential of AI while ensuring ethical, reliable, and transparent AI deployment together!

Seia

Understand the performance and impact of deployed ML models.

Understand the performance and impact of deployed ML models.