Infiheal Participated in the UK–India AI Collaborative Innovation Programme for Social Impact
Learn how Infiheal showcased innovation in inclusive AI mental healthcare through Healo at the AI for ALL Challenge.
Infiheal was selected among leading startups to participate in the UK–India AI Collaborative Innovation Programme for Social Impact, an initiative focused on shaping responsible and public-interest-driven AI. Representing the team, Nidhi Sehwag, Product Operations Manager at Infiheal, attended the two-day programme and engaged with entrepreneurs, researchers, and policy leaders in discussions on building AI systems that prioritize trust, accountability, and inclusion.
Why Responsible AI Needs Cross-Sector Collaboration
Conversations around responsible AI have largely stayed within silos. Technologists build models. Policymakers write frameworks. Clinicians raise concerns. Rarely are all three in the same room, working through the same problem at the same time.
This programme was designed to change that. Participants from very different professional domains were intentionally placed into cross-functional teams, so that the gaps that tend to stay hidden inside individual disciplines could surface through genuine collaboration. For Nidhi, being in that room meant exchanging perspectives with entrepreneurs, researchers, and leaders who are all building toward the same goal of AI that serves the public interest, but coming at it from entirely different angles.
The Problem the Sessions Surfaced
One issue came up consistently across working sessions: high-quality therapeutic datasets do not exist at scale, and the field has not found a credible way to address that.
What currently exists falls short. Paid access to real therapy transcripts raises serious ethical and consent concerns. AI-client conversations sold as training data are largely unstandardised. Synthetic data is growing in use, but has no reliable quality benchmark that the field can stand behind. This is not a gap that better models alone can close. It sits at the foundation of everything built on top of it.
What Infiheal Explored During the Programme
The cross-functional format brought Infiheal's work into direct collaboration with Dr Anirudh Anilkumar, who was independently developing an AI client to simulate complex therapy cases for therapist training, and Rama Krishna B. Working together, a shared question took shape: can high-quality therapeutic training data be generated without involving real patients, and can it genuinely be trusted?
The approach paired an AI therapist with an AI client simulating diverse psychological profiles. Generating conversations was the straightforward part. Evaluating them rigorously was not. To address this, the team built an LLM-based Dataset Judge that scores generated conversations across four criteria: empathy, safety, therapeutic framing, and emotional validation. Only conversations clearing all four were accepted.
To avoid circular validation, synthetic conversations were benchmarked against real-therapist conversations using the same judge as a reference standard. The early proof of concept was small, five to six conversations, three of which passed. The point was never the volume. It was demonstrating that a replicable, ethically grounded method for generating and validating therapeutic data could exist, one that reduces cost, time, and ethical risk, and builds shared infrastructure for the field at large.
Rooted in Infiheal's Broader Mission
Infiheal's founders, Srishti Srivastava and Utkarsh Srivastava, have built the company around clinically validated AI for mental healthcare within responsible AI frameworks, with safety and trust as non-negotiables. The company has been recognised by Prime Minister Narendra Modi in Mann Ki Baat and has presented its work at the World Economic Forum in Davos.
Participation in this programme is consistent with that trajectory. Responsible AI in mental health is not only a product question. It requires engagement at the level of policy, governance, and foundational research, and that is exactly the kind of work this programme was built to advance.
More about Healo can be learnt here: https://infiheal.com/ai-mental-health-chatbot
More about The Dialogue can be learnt here: https://thedialogue.org.in/
