I completed the MIT Professional Education Chief Technology Officer Program on October 2, 2026. As a member of Group 2, I worked on MANAS, a proposed AI companion for mental wellness awareness and support.

The underlying platform, MAANAS, stands for Multimodal AI for Awareness, Neurocognitive Analysis & Support. Our project connected a social challenge with technology strategy, system design, investment planning, and responsible AI evaluation.

The Problem and Proposed Solution

Our team examined gaps in recognizing distress, seeking help, accessing care, and maintaining ongoing support.

MANAS proposes combining consented text and voice check-ins with optional video and permitted health signals to support awareness, provide supportive dialogue, and connect users with human assistance. The final pitch includes prototype screens with fictional data. Clinical benefits remain to be validated.

1. Building the Vision

We established the project’s purpose, intended users, and guiding values, including empathy, dignity, and privacy. The problem-definition work examined recognition, acceptance, action, system, and continuity gaps.

This stage helped us translate a broad concern into a proposed product direction and identify where technology could support existing care pathways.

Supporting documents: Build a Vision; Problem Statement Resubmission—Draft.

2. Modeling the Solution and Roadmap

We developed an Object-Process Model and an expanded AI model to describe the proposed system and its interactions.

Our broader roadmap organized development into four phases: research foundation, prototype and pilot, scale and integration, and longer-term expansion. The final pitch made progression conditional on safety evidence and partner readiness.

Supporting document: Model a Roadmap of Your Solution.

3. Planning the R&D Portfolio

We divided the proposed development effort into five workstreams:

• Signal discovery, requirements, and regulatory considerations.
• The MVP platform and pilot application.
• Multimodal detection models and evaluation.
• Privacy, safety, and governance.
• Pilot programs and measurement dashboards.

The portfolio assigned budgets, schedules, and risks to these workstreams. It proposed a base investment of $13.2 million, plus a 15% management reserve, for a rounded total of $15.2 million. These figures are planning estimates, not secured funding.

Supporting document: Research and Development Portfolio.

4. Applying the Four Phases of Machine Learning

We examined the MANAS concept through four phases:

• Understanding data: Exploring voice, video, text, and behavioral features.
• Making predictions: Considering regression, classification, and temporal models.
• Making decisions: Comparing decision frameworks based on available information and changing user states.
• Evaluating causality: Proposing comparative studies to test whether an intervention produces meaningful benefit.

This work distinguished detecting a pattern from demonstrating improved outcomes. The materials describe an evaluation proposal; they do not report completed clinical trials.

Supporting document: Four Phases of Machine Learning.

5. Optimizing the Generative AI Strategy

We compared six approaches across technical capability, integration, cost, privacy, and ethical or regulatory fit.

The proposed architecture combines local processing with routed hosted models. It includes a supervisor, input and output screening, a deterministic crisis-response path, and clinician review.

The recommendation depends on appropriate agreements, safety evaluation, and staged rollout. Vendor capabilities, pricing, and regulatory assumptions require confirmation before implementation.

Supporting document: Optimize Your AI Strategy.

6. Presenting the Final Solution

Our final pitch brought together the user journey, proposed architecture, roadmap, institutional payer model, and investment requirements. It identified clinical partners and a pilot sponsor as priorities for the next stage.

Earlier recognition, greater reach, improved retention, and fewer relapses remain hypotheses for validation. The pitch proposes measurement against a comparison group and ongoing safety monitoring.

Supporting document: Pitch Your Solution.

What This Work Demonstrates

The MANAS project gave me an opportunity to connect technology leadership with concrete decisions about scope, architecture, investment, and evidence.

These deliverables represent collaborative work by MIT CTO Program Group 2. Together, they document the development of a proposal and prototype, with further validation needed to establish clinical benefit.

Explore the Project

Visit MANAS