Clearhead Improves Access to Mental Health Support with Enhanced AI Capabilities on AWS

Auckland-based startup Clearhead was founded to transform access to proactive mental health support. Established in 2018, the company has created a solution that adapts to its users, offering a holistic approach to recognizing mental health issues and supporting individuals throughout their mental well-being journey. As a leading employee assistance programs (EAP) provider across New Zealand and Australia, Clearhead’s wellbeing platform provides interactive AI tools to develop emotional intelligence, a mental health resource library, access to over 1000 therapists and psychologists for in-person or online therapy, and comprehensive reporting of psychosocial hazards in the workplace.

Within the app, users can also access a chatbot that serves as a digital therapist, supporting increased self-awareness regarding their mental health. This feature utilizes generative artificial intelligence (AI) to customize interactions, providing timely and personalized support for each user interaction.

Clearhead wanted to improve chatbot responses through better contextualization—incorporating all previous interactions with the wellbeing platform, not just chats. The company also hoped to integrate generative AI technology with other platform features such as therapy booking, making it easier for users to be matched with the right therapist. However, Clearhead found that its existing AI solution couldn't meet these goals. Training and operating its large language model proved costly, and keeping the model up to date was challenging, leading to chat users sometimes waiting 10 seconds or more to receive a response.

To address these challenges, Clearhead turned to Amazon Web Services (AWS) and participated in the AWS ANZ Generative AI Accelerator program, which offers startups support and resources to build and refine their AI solutions. Clearhead's participation in the accelerator program, part of AWS Startup Loft, facilitated the adoption of Amazon Bedrock, a fully managed service that offers a range of high-performing foundation models (FMs) to streamline the development and scaling of generative AI applications.

With Amazon Bedrock, Clearhead improved generative AI performance, adding cost-effective text summarization to its existing machine learning models. As the Clearhead team continued to meet weekly with its AWS Solutions Architects, it also integrated Amazon Titan, a family of FMs pretrained by AWS on large datasets, with its existing machine learning models. “We would highlight if we needed any support in which areas, and AWS Solutions Architects promptly addressed our needs, providing the necessary information. The process progressed swiftly," says Connolly.

Within seven weeks, Clearhead successfully implemented Amazon Bedrock with Amazon Titan, completed a rigorous quality assurance process and transitioned workloads into production. Clearhead employs Amazon Virtual Private Cloud (Amazon VPC) for full control over its virtual networking environment, including resource placement, connectivity, and security. This approach ensured a secure connection between Clearhead's Amazon VPC and Amazon Bedrock, safeguarding the company’s data and giving customers peace of mind.

In addition to addressing Clearhead's data security and compliance requirements, the fully managed Amazon Bedrock solution helped the company resolve cost concerns through its pay-as-you-go model. States Michael Connolly, chief technology officer at Clearhead, "By using AWS, we pay just a few hundred dollars for generative AI, as opposed to the thousands of dollars required to maintain underlying hardware."

Furthermore, Connolly and his team gained more time to innovate because they no longer had to manage the AI solution’s infrastructure. The time saved is used to test other FMs available through Amazon Bedrock—such as Anthropic Claude, which writes long-form documents. “Amazon Bedrock gives us the opportunity to find the best FMs for our different use cases,” says Connolly. “Through its management console, we can swap out models very quickly with minimal work and experiment a lot faster.”