Building Investor-Ready AI Mental Health Assistant MVP on AWS Bedrock for Startup
Achievements:
Summarize this case study with AI:
Contents
- Client Profile
- Introduction
- Background
- The Challenge
- Our Approach
- Results
- Client Feedback
- Conclusion & Next Steps
- FAQ
Client Profile
Company Name: i Got This! Mental Health and Wellbeing Assistants
Founded: 2024
Headquarters: United Kingdom
Industry: Digital Health / AI Avatars / Mental Health Technology
Founder: Ann Watson (solo founder; former head teacher, carer, charity and local government experience)
Funding: Self-funded, supported by AWS Activate
Company Size: Early-stage startup (solo founder with Perfsys as technology partner)
Website: igt-pa.co.uk
Crowdfunding Campaign: Crowdfunder Profile
Introduction
i Got This! is pioneering AI-powered mental health assistants designed to listen, respond with empathy, and guide users through calming techniques when support is needed most. Unlike purely clinical tools, these assistants fill the gap between therapy sessions and the moments when people hesitate to reach out, with the specific aim of prevention of self harm and suicide.
Perfsys partnered with founder Ann Watson to create an investor-ready MVP that balances user safety, cutting-edge AI, data security, and scalability on AWS. By combining empathetic avatars with a secure serverless backend, we helped transform a vision into a demo-ready product capable of attracting investor attention.
Background
Ann Watson founded i Got This! in 2024, driven by decades of personal and professional experience supporting people in challenging circumstances. As a carer for her autistic daughter, former head teacher, and advocate for families navigating complex systems, she saw first-hand how rising mental health challenges impact not only those who are suffering but also their families and society as a whole.
Her idea was simple but ambitious: build an accessible, AI-driven assistant that listens without judgment, promotes healthier habits, and guides users toward professional resources when needed. After a failed attempt with another vendor on a mobile-first build, Ann turned to Perfsys in April 2025. By July 2025, we had signed and launched the project with a web-first, mobile-friendly MVP, giving her a working product faster and at lower cost.
The Challenge
The project came with a few big challenges. Because Ann’s solution introduces a new way to combine AI and human empathy, protecting intellectual property and data security was a key part of our design approach. The client wanted to use avatars to create a human-like experience, but testing on the original third party platform was proving very expensive. At the same time, the MVP still needed to impress investors — it had to be scalable, secure, and built to follow Government guidelines as well as AWS best practices.
Technically, this meant:
- Decoupling avatar UI from AI logic
- Designing a secure, real-time communication layer on AWS to connect the avatars with the AI engine
- Implementing guardrails for sensitive topics like self-harm and suicide using AWS Bedrock
- Allowing the AWS Bedrock AI Agent to return structured data, like actions (e.g., redirecting to an exercise page)
- Ensuring a mobile-friendly MVP with a path to native mobile (React Native) later
Our Approach
We structured development in three clear phases.
Phase 1 — SDK Validation
We began by experimenting with third party avatars’ SDK to understand capabilities and constraints. A lightweight UI-only app confirmed feasibility and provided the foundation for the next steps.
Phase 2 — Web App MVP on AWS Amplify
We delivered a web app hosted on AWS Amplify, with Cognito for secure sign-up/sign-in and profile management. This enabled early user testing where the third party avatar handled conversations through its built-in AI.
- Authentication and secure user pools via Cognito
- Personal information captured during onboarding
- Initial conversational flows (e.g., breathing exercises) handled by third party AI engine
Avatar Integration in the AWS Cloud using AWS Amplify framework and AWS Cognito authorization.
Phase 3 — Decoupled AI on AWS Bedrock
To reduce cost and gain control, we separated avatar UX from AI logic. Avatars became the interface only, while AI processing shifted to Amazon Bedrock.
- Backend: AWS Lambda and API Gateway power the interaction between the avatars and the AI engine, ensuring performance and scalability.
- Communication layer: The backend securely manages real-time interactions so that avatars respond naturally while the intelligence runs on AWS Bedrock.
- AI agent in Bedrock: System prompt for role and ethics, guardrails for sensitive topics, and structured JSON outputs for actions.
- Personalization: User profile data captured in Cognito onboarding was passed as session metadata into Bedrock with every request, enabling more tailored responses.
- Future extensibility: The design leaves room to connect a curated Knowledge Base in Bedrock, so the agent can access relevant mental health content beyond its system prompt.
Results
By the end of Phase 3, the MVP met both the technical and business goals:
- >90% cost savings: Avatar testing usage dropped from ≈300–350 minutes/month to ≈20 minutes by moving AI logic to Bedrock.
- Scalable AWS stack: 7 AWS services (Amplify, Cognito, API Gateway, Lambda, DynamoDB, Bedrock, S3/CloudFront) built a due-diligence-friendly, serverless foundation.
- Bedrock control & safety: Hosting AI on Bedrock gave us full control over prompts, metadata, and guardrails, ensuring conversations were safe, personalized, and extensible. It also opened a path to add a Knowledge Base and experiment with hybrid model routing for performance.
These outcomes positioned i Got This! as investor-ready — not just a prototype, but a secure, compliant, and scalable product.
Thinking about building your own AI agent or assistant?
At Perfsys, we help startups turn ambitious ideas into secure, scalable MVPs that investors can believe in — without wasting time or money on the wrong technology. Contact us or Book a call with specialist
Client Feedback
The founder shared their experience in a verified Clutch review:
“Perfsys were extremely professional on every level. Their interest in the project made me feel they were my own in-house development team. They listened, understood the priorities, and delivered everything exactly as requested.” — Ann Watson, Founder, i Got This! iGT-PA Ltd (Clutch Review, Oct 2025)
The client awarded Perfsys 5.0 out of 5 across all categories — quality, schedule, cost, and willingness to refer.
Conclusion & Next Steps
The MVP proved the vision: empathetic AI avatars, powered by AWS Bedrock, ready to support mental health at scale. Our ongoing work with i Got This! focuses on three fronts:
- Performance: exploring hybrid model routing (fast models for small talk, richer models for complex flows).
- Knowledge: building a curated knowledge base to make the agent smarter beyond system prompts.
- UX: onboarding improvements and avatar personalization ahead of upcoming showcases.
FAQ
What is Amazon Bedrock?
Amazon Bedrock (AWS Bedrock) is a fully managed service that lets you build and scale generative AI applications using foundation models from providers like Amazon, Anthropic, AI21 Labs, Cohere, and Stability AI—all through a single API. It removes the need to manage infrastructure, offers tools for customization and safety guardrails, and integrates seamlessly with other AWS services. With flexible pricing and serverless scalability, Bedrock makes it easier for businesses to develop secure, enterprise-grade AI solutions.
How do third party avatars integrate with AWS Bedrock?
Through secure, real-time integration built on AWS. The backend manages continuous interactions between avatars and the Bedrock AI engine, ensuring reliable communication and safety features.
Why not use avatars built-in AI?
Testing minutes are expensive and limited. By moving AI to Bedrock, we reduced dependency, gained flexibility, and cut costs. This approach also gave us full control over prompts, metadata, and guardrails, making the AI agent safer and more adaptable.
Which AWS services were used to build the MVP?
The MVP relies on a modern serverless stack: Amplify, Cognito, API Gateway, Lambda, DynamoDB, Bedrock, S3, and CloudFront. Together, these services ensure scalability, reliability, and compliance with AWS Well-Architected Framework best practices.
How is user safety handled in an AI mental health assistant?
Guardrails in Bedrock plus ethical system prompts ensure the assistant avoids clinical claims and instead signposts to professional help when needed. Sensitive topics like self-harm are recognized, and the AI agent redirects users to appropriate resources rather than attempting to diagnose.
What makes this MVP investor-ready?
The architecture follows AWS Well-Architected Framework principles, including security, cost optimization, and performance efficiency. By hosting intelligence on Amazon Bedrock, the team reduced costs, improved flexibility, and created a product that is easier for investors to evaluate and scale.
How does Perfsys approach MVP development on AWS?
Our team specializes in building MVPs on AWS. We combine rapid delivery with secure, scalable architecture. For AI projects, we focus on cost-effective experimentation, clear separation of concerns, and a serverless foundation that can grow into a production-ready product.
Can Bedrock-based AI agents be personalized for each user?
Yes. User profile data captured through Cognito is passed into Bedrock as session metadata, enabling responses tailored to each individual. This allows the AI assistant to adapt recommendations, such as breathing exercises or reminders, to the specific user context.
How do AI Agent Development Services from Perfsys support startups?
Perfsys offers AI Agent Development Services that help startups design, build, and scale intelligent assistants. From integrating empathetic avatars to setting up secure AWS Bedrock agents, we provide end-to-end support so founders can focus on vision while we handle technical complexity.
What future improvements are planned for i Got This!?
The roadmap includes faster performance through hybrid model routing, expanding knowledge with a curated Bedrock Knowledge Base, and improved UX with onboarding and avatar personalization. These steps will make the AI agent even smarter, safer, and more engaging.
How does this case study relate to other Perfsys projects?
Just like with i Got This!, we help startups across industries cut costs, scale faster, and impress investors. Whether it’s healthcare, fintech, or SaaS, Perfsys applies the same principles: cloud-native architecture, AI integration, and business-focused delivery.
Ready to grow your startup on AWS?
We help early-stage teams scale fast and affordably with AWS credits, serverless architecture, and expert support.
Other case studies
