04/02/2026
π¬ βWhen choosing how to apply AI models to modern challenges, there is no predictable answer. We must, to borrow a technique from social care, review our impact, continuously improve and remain connected to lived experience.β - Sudha Regmi, Director of Data & AI
β οΈ Safety and ethics must be at the forefront of applied AI, especially in health and social care
π Anthropic's newly published constitution for Claude puts it plainly: safe β ethical β compliant β helpful, with hard constraints for high-stakes use. That ordering recognises that how AI is built is just as important as what it produces
π» At Nourish, this deeply resonates with our own approach to applied AI in health and social care, with our focus on
π Incentivising the right kind of outputs for the context
π¨βπ Human-in-the-loop validation with clinical experts
π Tight feedback loops with our customers to pressure-test what βgoodβ looks like in practice
π€ As our Director of Data & AI, Sudha Regmi, reflects: "applied AI isnβt about abstract alignment, itβs about continuously reviewing impact, learning from lived experience, and earning confidence through process, not black boxes."
π Read the full blog to explore the parallels and Nourishβs approach to applied AI in care
π https://eu1.hubs.ly/H0rttTm0
Our Director of Data and AI Sudha Regmi explores the new Claude constitution and the importance of human principals in applied AI design.