AI Coach for Mental Resilience

The problem is a double whammy: a national mental health crisis in India, and exacerbated with the deep marginalization that victims and survivors of modern slavery face in addition. 

The mental health crisis in India is acute. 25% of India’s population suffers from common mental health disorders, such as depression, stress, suicidal risk. Suicide is the leading cause of death among 15-29-year old Indians. And care is disproportionately limited. 85% of those affected with a common mental disorder did not reach out or access professional help. For every 100K population, India only has ~0.75 Psychiatrist available (vs recommended 3), and ~0.07 Psychologist vs >30 (in the US). 

 The pain points are exceptionally acute for the excluded and underserved. An estimated 11 million individuals are trapped in various forms of modern slavery in India, the country with the highest victims globally. This manifests as bonded labor, forced labor, child labor, human trafficking for sex and domestic servitude, and forced marriage - and disproportionately affects underprivileged populations such as women, lower-caste, poor individuals. Research by IIED indicates that in India, Dalit households, facing severe discrimination, are three times more likely to migrate than higher caste households, which exacerbates their vulnerability to modern slavery. Despite stringent laws, child labor remains a significant issue, with minors employed in hazardous industries.

Recent findings suggest that the Indian government, nonprofits and other stakeholders are able to identify and support only 31% of victims of modern slavery. The NCRB report highlighted that among the rescued trafficking victims in India, 80% reported severe anxiety, 75% reported depression, and 65% exhibited symptoms of PTSD. Survivors also face stigmatization, lack of access to education, and difficulties in obtaining employment. This lack of access to critical resources and services not only affects survivors’ well-being but it also increases the risk of getting re-trafficked. In addition, while support organizations play a crucial role in aiding survivors, they encounter challenges such as limited reach, and sometimes a lack of coordinated efforts across different sectors. This results in lack of utilization of available resources and inefficiencies in delivering comprehensive care.

Wysa is a conversational AI chatbot. It acts as a personal AI coach for mental resilience, where users can stay anonymous. 

Wysa is offered multi-modally, on an app on Android or iOS, or on the web. It listens empathetically, gives people a non-judgemental & zero-stigma space to explore their emotions, and promotes internal reflection. By offering intrapersonal skills, the AI coach creates psychological safety and has proven to improve resilience skills, lower stress, reduce symptoms of anxiety and depression, and prevent suicides. 

As an AI powered solution Wysa is able to fulfill unmet needs of vulnerable communities. It offers anonymity and unrestricted access, users can call upon Wysa at 3 AM in the night and speak without fear or associated stigma. Research indicates that people open to Wysa’s AI 3x faster than they do to a human therapist. Through AI, Wysa offers consistent quality and experience to all its users at all times. It can also detect patterns in conversation that may indicate distress or suicidal ideation more consistently than humans, who may suffer from fatigue or bias. Wysa has been able to save over 400 lives through detection and prevention. This feature is especially relevant in the context of modern slavery survivors who may experience acute stress or panic attacks at unpredictable times and need immediate access to support. In addition, being AI-enabled makes our solution scalable across the world ensuring quality care is available to all survivors who need it from sex trafficking victims in Romania to forced labor in Bangladesh. 

For the most vulnerable people in India, such as survivors of modern slavery, this is their first experience in feeling heard and supported without judgment. The AI Coach then helps them build skills for resilience such as cognitive reframing, grounding, and support in improving external relationships, behaviors and taking actions for change. Our experience with diverse and underserved populations, including rural women, adolescents, and low-income slum dwellers, has shown positive outcomes, demonstrating Wysa’s ability to adapt and deliver significant benefits across different societal segments. Our solution is able to contextualize its support, using AI models that can detect and identify user context and messages.  The solution is available not only in English, but we have also built the AI Coach in Hindi (most spoken Indian language) and Marathi (3rd most spoken Indian language) - and is slated to be multilingual in other regional languages within the next year.

Wysa also acts seamlessly as a triaging platform - that can screen, assess, and signpost users to any existing resources such as specialized mental health services, legal assistance. Importantly, Wysa’s AI includes a robust SOS management protocol. With vulnerable populations, Wysa is able to detect self-harm or suicidal ideation and offer safety plans, grounding techniques, a warm hand-off to local and/or national helplines, and customise any escalation pathway.

Over the past 8 years, our AI models have undergone more than 700 rounds of improvements. Today, Wysa is built to the highest global standards of security, privacy, ethics, and clinical safety. We are leading the way in using AI in a responsible and ethical approach, built on the following practices: anonymity, privacy, deterministic, legitimate use and fairness, infosec, and clinical safety. 

i) Anonymity: Given the sensitive nature of the data involved, assurance of anonymity is crucial to fostering a safe and trusting environment. First and foremost, Wysa requires minimal personal information from the users, ideally just a username that does not have to be linked to their real identity. Furthermore, the handling, analysis, and sharing of data are conducted in such a manner that it becomes impossible to link any collected information back to an individual. This involves sophisticated data encryption, the use of anonymized identifiers, and strict controls over data access. 

ii) Privacy: Explicit consent from users, transparent data handling policies, no storage of data on OpenAI for training, redaction of any personally identifiable information (PII), and adherence to international privacy standards like GDPR.

iii) Deterministic: Establish repeatable and strict deterministic protocols, to ensure AI systems perform consistently across various scenarios, minimizing the risk of unexpected outcomes & hallucinations. Structured responses that adhere to predefined patterns are essential for maintaining the integrity and trustworthiness of AI interactions in sensitive applications.

iv) Legitimate Use and Fairness: Proven solution that lowers mental health risk. In addition to being effective, we ensure our solution is equitable. Our AI is co-designed with users (esp. low-resource, vulnerable populations) to ensure our algorithms do not perpetuate existing biases or inequalities. The development process is designed to be inclusive and fair. 

v) Information Security: Exhaustive compliance protocols aligning with ISO 27001 standards. We employ a detailed 53-item compliance checklist that covers everything from data encryption to regular security audits and vulnerability assessments. This ensures all potential security risks are addressed proactively.

vi) Clinical Safety: Adherence to stringent clinical safety standards, such as DCB0129 to manage and mitigate any AI risks in healthcare settings. Our protocols for handling potentially harmful cases are carefully crafted and regularly reviewed to ensure they are up to date with current clinical guidelines and best practices. Repeatable assessment for following protocol, handling potentially harmful cases without offering medical advice.

Our solution serves people who have no access to care - this includes marginalized / vulnerable communities, given their amplified burden of stigma, judgment, and lack of resources. We address the unique needs of survivors, especially adolescents, women, and migrant workers. See below demographic data for our India user base:

• Gender. ~70% are female and ~30% male. 

• Age. More than ~50% of our users are between the ages of 18-24, another 25% are between 25-34 years, and the remaining go upwards of ages 65+. 

• Cities. In the Top 30 cities by # of users, we have 9 cities that are Tier-I, 15 cities that are Tier-II, and 5 cities that are Tier-III.

• Income. Android accounts for 90% of downloads vs 10% for iOS gives us low-income proxy data. Note: Android users are more likely than iOS to be younger, lower income and living in rural/semi-urban areas: for example, see here.

As an example: a recent project work with survivors (voluntarily self-identified) was part of an enterprise pilot with a blue-collar skilling organization, that works with migrant workers and rehabilitates them from source states to manufacturing city hubs. In a month, we provided the Wysa to 150 workers. 80% men, 85% between 13-25 years of age, and 40% earning less than 100 USD in a month. 

Our solution is useful for under-served communities, who have very specific needs: Those who need anonymity: Psychologically, we facilitate rehabilitation through AI counseling and resilience skilling, by being a 24x7, fully anonymous safe space to help survivors overcome trauma and regain strength. Emotional and psychological healing is a crucial step in the overall recovery model for survivors. 

Those who need holistic support: Recovery of survivors enables their mental, emotional, and physical well-being. Stress and trauma manifests physically as sleep disturbances, fatigue, and somatic symptoms. Wysa supports these directly through techniques like CBT, mindfulness, grounding; and indirectly by psychological healing leading to improved physical health. Further, skills such nurturing relationships, negotiation, managing conflict, etc. help improve overall quality of life, enabling survivors to thrive in society.

Those who need interconnected (not point) solutions: We connect and triage to existing resources, which can be customized for different contexts. This may include legal aid, schemes for financial assistance, and support groups for skilling and employment. SOS and crisis management is a particularly important resource connection that our AI Coach can identify and escalate.

Finally, we improve the ecosystem through crucial data insights. We gather anonymized, secure data at a cohort / population level including user behavior, topics of concern, and longitudinal screening outcomes. This allows us to identify patterns and outcomes across different demographics and intervention strategies. Key stakeholders in the ecosystem can leverage these insights to improve the reintegration process for survivors, and contribute to broader systemic changes.

The numbers support the impact too: Wysa has a 90% helpfulness rating from users. Wysa has proven in research a 30% reduction in depression in 2 weeks, 40% reduction in anxiety in 8 weeks, and 80% increase in employee resilience over 8 weeks.

Wysa began not merely as a business venture but as a deeply personal mission to address mental health challenges, inspired by our founders' personal experiences with depression, caregiving, and suicide. This profound personal connection informs our understanding and sensitivity towards the complexities of mental health, particularly for the less privileged without access to any help.

Our Team Lead, Rhea Yadav, has worked closely with survivors of modern slavery before she joined Wysa. She worked with Freedom Rising, and led the development and implementation of the Freedom Rising Leadership Program for hundreds of survivors of modern slavery in Tamil Nadu, India. This hotspot primarily included survivors of bonded labor from the textile industry and brick kilns in South India. Rhea has spent a full year doing field work with the community alongside the local hotspot team, to understand the inner resilience journey, and then design a residential program for nurturing the next generation of survivor frontline leaders.  Mental resilience was then a core part of survivorship, and today at Wysa - she extends the same principles to expand access to care for all survivors in India, and then around the world. 

Our leadership team also includes founders - Jo Aggarwal, our CEO, who brings over three decades of experience in skilling, pioneering elearning and innovation, especially in the conflict-ridden Middle East during the start of the Arab Spring, and Ramakant Vempati, our President, who offers extensive experience in development and impact financing, microenterprise support, and consulting. Smriti Joshi, our Chief Psychologist, is a licensed clinical psychologist with over two decades of experience working with all kinds of survivors including runaway children, trafficking survivors, and sex workers. 

In the last year, we have also instituted this approach for Indian users, especially survivors and vulnerable populations, by setting up co-design labs. These work as centers where a local partner works alongside us to design the solution, customize and contextualize, all with real users themselves. We already have one co-design lab with Udaan Center for Kids in Maharashtra, which works with young people living with Type-1 Diabetes and come from a rural, and low socioeconomic background. With them, we’ve bee able to build the Wysa AI Coach in Marathi language. We are working to institute two more co-design labs in Karnataka and Andhra Pradesh this year. 

Overall, Wysa has always been user-led. We started the first design of a chatbot conversational flow 8 years ago with users out of a rural clinic in Aurangabad (tier 3 city in India). Today, we’re working towards forging partnerships with the government in India, and are in advanced conversations with the State Govt. of Maharashtra to deploy Wysa AI Coach for all adolescent girls. Being able to operate at scale, in collaboration with ecosystem players, embedding ourselves in the government channels, and continuing to keep the user at the center - this is the DNA our team has.

Wysa AI Coach has provided access at scale, shown high adoption, and usability.  

• Used by 1 million Indians, and 6.5 million total users across 95+ countries

• 750 million conversations

• 60+ enterprise partners

• Large public health rollouts in MOH Singapore, NHS UK, and Multiple US States (Maharashtra State Govt in India soon)

• 4x engagement, 10x access to mental health

Wysa has the most evidence of any mental health app globally. Clinically validated with over 40 peer reviewed publications and clinical trials on clinical improvements in both mental health (PHQ9, GAD7, Stress Resilience) and physical health (Pain Interference and Physical Function). Select research partners include Washington University at St Louis, Cambridge, Columbia, NHS UK, Oxford, LSE, NIH. 

It is proven to work across cultures, demographics, and socioeconomic classes. And address all types of stress and distress, including relationships, financial, legal, health, trauma, and abuse. It has proven best-in-class outcomes: Evidence of Therapeutic Alliance comparable to Human CBT, 30% reduction in depression in 2 weeks, 40% reduction in anxiety in 8 weeks, and 80% increase in employee resilience over 8 weeks.

Wysa is recognized as a top innovation globally. Recipient of the FDA Breakthrough Device Designation, NHS AI Award, Mozilla Best of Privacy, World Economic Forum Youth Mental Health Winner.

With the AI Coach built in Hindi over the last year, we have started to see similar outcomes from pilot deployments with marginalized populations across migrant workers and loneliness, rural women and sexual health concerns, and young people and marginalized queer identities. 

We are looking to receive funding to build, expand, and support the integration of our eTriage function in connecting survivors to existing resources. Wysa offers a comprehensive continuity of care already, but survivors require a more holistic triaging of resources, with lived experience of communities - and we’re looking to do that in a more concerted manner. We are hoping to gain connections with fellow innovators, forge partnerships with key players in the ecosystem for survivor support, and create networks with survivor communities to strengthen our co-design labs. 

The Wysa AI Coach listens empathetically, gives people a non-judgemental space to explore their emotions, and creates new mental resilience skills with just 10 mins of chat everyday. 

• Available 24/7. Real-time unlimited support whenever it’s needed. Some people turn to Wysa 7 times a day.

• 100% Anonymous. For user data privacy and security. Instantly removes stigma. 

• Responsible AI. Deterministic AI with Natural language understanding (NLU) empathises, then guides to the right techniques.

• Clinically validated. Built with clinicians, as first step of care. Most evidence of any mental health app. 

• Meet users where they are. Available in English, Hindi, Spanish & multi-lingual in 6 months. Across platforms: app, web, whatsapp, voice 

Two specific things that are innovative:

1. User engagement: Wysa stands out by allowing free text input for 80% of prompts during interactions, compared to just 20% with any competitor. Our patented technology identifies when the therapeutic alliance might be weakening, potentially signaling user disengagement. This allows for timely interventions to re-engage and refocus the user, contributing to why 91% of our users find Wysa helpful.

2. Continuity of care: Wysa offers a comprehensive approach by covering multiple aspects of the support pathway, including eTriage as part of the digital ‘front door’, supporting users on waitlists, serving as a digital therapeutic in clinical programs, and enabling remote patient monitoring. We are expanding upon the eTriage function and incorporating other relevant stakeholders - legal, philanthropic and government, improving awareness and catalyzing broader positive impact from other ecosystem players as well. 

People open to Wysa’s AI 3x faster than a therapist. This level of insight into the inner lives of vulnerable people is a game-changer for creating large-scale behaviour change. It is this effectiveness, scalability to virtually any smartphone, and clinical efficacy - that makes the Wysa AI Coach unique in supporting resilience and improving lives.

Wysa supports survivors both internally (inward, intrapersonal, resilience, skills) and externally (information asymmetry, matchmaking of resources, utilization). 

Internally. Our solution is designed to address the complex challenges faced by survivors of modern slavery by utilizing the INTRApersonal model for skilling and resilience building. This makes the ‘self’ a catalyzer for change, and strategically supports a survivor’s journey towards recovery and reintegration into society:

1. Internal Reflection (I) and Non-judgmental Listening (N): Identify limiting beliefs and thoughts. Reflect on unhelpful thoughts, feelings & behaviors. Listen to thoughts & feelings without judgment. Promoting self acceptance.

2. Thought Shaping (T): Cognitive and behavioral techniques. Challenge unhelpful thoughts & beliefs. Develop new, more constructive thought patterns. Build self awareness, confidence & resilience.

3. Relationship Skills (R): Examination of relationships. Encourage effective communication & trust. Create stronger, healthy connections. Improve emotional wellbeing.

4. Actioning Behaviors (A): Set specific, achievable goals to provide direction & motivation. Develop effective coping strategies & problem- solving skills. Encourage consistent practice & application. Monitor progress & adjust approaches.

Externally. In addition to the self, Wysa also creates impact by ensuring survivors are more informed and equipped to access support and resources. 

• Wysa will support survivors through a step-by-step process that begins by matching critical resources to individual needs. Using its advanced triage system, Wysa evaluates the specific needs of each survivor, ensuring they receive targeted support, whether it's legal aid, educational opportunities, or job placements. 

• Next, Wysa informs survivors about these resources, their benefits and providing guidance on how to effectively utilize them. For instance, our tool around ‘Planning Difficult Conversations’ was one of the most used (1/3rd of the target group) and helpful tools for vulnerable rural women in addressing their sexual and reproductive health concerns. 

• Finally, Wysa's digital front door function provides immediate, easy access to these resources, eliminating information asymmetry and empowering survivors towards rebuilding their lives. In our digital front door function with a large enterprise, we were able to connect 162,000 users to the right care and achieved a 98% satisfaction score for this function.

By embedding Wysa's specific interventions within the INTRA framework, it becomes clear how each element of the service contributes to a comprehensive support structure for survivors, empowering them to take control of their mental health and successfully reintegrate into society.

We want to serve 10 million users in India in the next 5 years. With at least 25% of these users from marginalized and vulnerable populations including survivors of modern slavery. We want the Wysa AI Coach to be embedded in the government channels in at least 2 States in India. And we want to ensure we maintain our benchmarks on adoption, engagement, and helpfulness as we scale. 

We are measuring progress towards this with the following framework:

1. Is the solution available where and when users need it?

   1. Users reached

2. Do users find it, and are they trying it? 

   1. Downloads, adoption

3. Do users use it enough?

   1. ###p#< users returning
   2. ###p#< of returning users having 5+ sessions or more (stickiness)

   3. Retention rate

   4. Engagement: Average session time and messages exchanged

4. Do users find value when they try it?

   1. User helpfulness score

   2. Qualitative feedback

   3. Topics, toolpacks, emotions

   4. Completion rates

5. Does this show outcomes?

   1. Changes in health outcomes

   2. Changes in other assessment scores

   3. Triaging matches and follow-ups

• At the core of the Wysa Solution are Artificial Intelligence models that use natural language processing (NLP), natural language understanding (NLU), machine learning, and deep learning to interpret the user's messages and enable the chatbot to listen to the user like a friend. 

• The platform utilizes over 100 NLU models to accurately interpret user inputs and context. A sophisticated rules engine then analyzes this data to determine the most appropriate response, while a vast, clinician-curated content library ensures that responses are both empathetic, timely, and clinically safe. 

• Additionally, our AI models are continuously improved by regular updates through batched, clinician-oversight driven analytics. These leverage insights from Wysa data (the largest domain-specific corpus of mental health conversational content in the world). 

• To support this advanced technology stack, Wysa employs a robust cloud-based infrastructure designed for scalability. Wysa leverages REST API’s along with websockets for its offering. The architecture is aligned with service oriented architecture and all API’s are internal at this moment. 

• The aspects of incorporating Generative AI in a responsible and ethical way, while ensuring the highest standards of clinical safety and data privacy - are captured in Q6 above.

• Users access Wysa through a mobile application (android and iOS), a web application, a whatsapp-based chatbot, or via tech integrations like API/SDK integrations and deeplinks. 

Wysa has 119 full-time and 35 part-time consultants/interns/advisors. 90% of the staff is based out of India, and the remaining (mainly sales, market-facing roles) are split across London, UK and Boston, US. We have a rich interdisciplinary team, with 70% of our team being core to the solution (e.g. Tech, AI, Therapist, Product), and remaining 30% in market-facing or cross-functional functions such as Marketing, Sales, Compliance, Research.

Our interdisciplinary team includes clinicians, conversation designers, AI/ML engineers, data scientists, researchers, public health practitioners, tech developers, product managers, business development and partnership associates, and accountants and more.

8 years, developing the AI solution for the world. 2 years, co-designing and building especially for unserved and vulnerable populations in India.

Our team is interdisciplinary and stronger due to its diversity. Wysa has 60% women employees. 90% of the team members live in India, and are people of color. Our India team comes from more than 15 different states, and resides across tier-I, II, and III cities. Our team speaks more than 20+ languages. And more than a third of our people bring the richness of lived experience with mental health. The co-founders built Wysa out of personal experiences with depression and bipolar in the family. We also work with partners in the field representative of our end users, within co-design labs for human-centered design.

Our solution team has exceptional leaders, many of whom led the Wysa B2C app to 6m users and a 4.9app rating in 5 years. Leaders have worked in grassroots communities, bring design-thinking to problem solving, ensure intersectionality in conversation design. Backgrounds range from ex-founders, product leads, PhD candidates, mgmt. consultants, and psychologists. 

Wysa’s business model effectively merges different go-to-market approaches to ensure maximum impact in a sustainable model. 

1. B2C Model: At the core of our B2C model is the Wysa app, which offers AI-driven mental health support through an intuitive conversational interface. This platform caters directly to individuals seeking help for mental health issues. Users can access a variety of tools and sessions based on Cognitive Behavioral Therapy (CBT) anytime and anywhere. The app includes a freemium model where basic features are free, while more comprehensive support tools and resources are available through in-app purchases and subscriptions. 

2. B2B Model: Wysa’s B2B model focuses on partnering with enterprises to offer wellbeing and preventative mental health support into their existing health and employee benefits frameworks. These are licensed on a subscription basis, and may include development fees for customisations and integrations. Enterprises gain both wellbeing impact and economic benefits from improved employee wellness. Bespoke data analytics are a core component of the value prop.

3. B2G Model: We also work with government systems at a population level, such as with Singapore, or integrated into healthcare pathways, such as with NHS UK. Typically the government is the payor, and it operates akin to a B2B model. Oftentimes, like with State Govt. in India, this model is alongside co-funding from philanthropic money. 

4. Grants: To extend our reach to underserved populations, we also leverage philanthropic funding. This typically supports product development of specific programs, languages, and distribution for service delivery to marginalized populations. Impact measurement and expanding our social mission to make mental healthcare accessible for all is the thesis. We also focus on expanding our evidence base and healthcare and research grants also form a part of this. 

Wysa is a for-purpose social enterprise. Last year, our revenue grew by 50%, and a similar growth forecast puts us on a path to breakeven by 2025-26. In order to continue our efforts of expanding our impact to vulnerable populations, we will look to increase philanthropic grants by 25% next year, and continue to retain commercial revenue for operational costs. Wysa’s financial sustainability is upheld by its dual business model which combines our revenue generating business that enables us to build cutting-edge AI and attract the best tech talent; along with our philanthropic donors who support product development and service delivery in areas where a market doesn’t exist yet.  

The success of our financial model is evidenced by the significant traction we have gained within the enterprise sector, which currently constitutes 80% of our revenue. The for-profit revenue ensures that we can continue to invest in cutting-edge AI technology and expand our offerings while maintaining high-quality service delivery. Moreover, our ability to attract philanthropic funding and grants—for example, the ACT Grants for development, and additional support from entities like British International Investment and USAID—demonstrates confidence in our mission and operational model from impact-oriented donors. These funds are directed specifically towards expanding our impact without burdening the operational costs, allowing us to innovate and scale our impact initiatives effectively.