Offline AI for Empowering Underserved Teens (Pilot PoC)

Project summary

The project addresses the digital divide by focusing on offline-first technology, ensuring the AI runs smoothly on low-end smartphones (2GB-3GB RAM) without an internet connection.

By capturing the unique psychological archetypes of local youth, we aim to provide a private, non-judgmental mentor that helps teens navigate insecurities and grow into their best selves.

Unlike general AI, our solution is powered by a Small Language Model (SLM) fine-tuned on a massive, hyper-local dataset of 2,000+ teen surveys and 120 parent interviews conducted face-to-face in a single Pilot City.

What are this project's goals? How will you achieve them?

We aim to prove that AI can provide high-quality emotional support when trained on specific local behavioral patterns rather than generic global data. We will achieve this through intensive face-to-face fieldwork in our Pilot City to capture authentic dialects, social pressures, and parent-child dynamics.
We will achieve accessibility by using Quantized Small Language Models (SLMs) designed to run locally on-device, catering to teens in low-connectivity areas using budget smartphones.
By processing all data 100% offline, we eliminate the risk of sensitive personal data ever leaving the user’s device, creating a truly safe space for self-expression.

How will this funding be used?

$5,600 - Research & Fieldwork
$5,000 - AI Development
$1,200 - Administration & Documentation
$1,200 - Contigency ~10%

Our Minimum Funding will focus on a "Technical Lean" approach, scaling the research to under 1,000 teenagers to prove the core feasibility of an offline, privacy-first SLM on low-end hardware.

Full Funding enables a "High-Fidelity" launch, expanding the dataset to 2,000+ teenagers and 120 parents to capture a complete spectrum of psychological archetypes.

Who is on your team? What's your track record on similar projects?

This is our first independent initiative as a collective, we operate under a Strategic Alliance Model. We have secured advisory commitments from:

Professional Psychologists with years of experience in adolescent clinical behavior.
Senior IT Professionals specializing in mobile architecture and machine learning.
Our core team brings a lean, execution-focused mindset, prioritizing data density over geographical spread by focusing on a single Pilot City to ensure the PoC’s success.

What are the most likely causes and outcomes if this project fails?

Likely Causes of Failure:

Despite optimization, the model might still experience latency on extremely outdated chipsets.
The "personality" of the AI might not resonate with the teens if the cultural nuances are lost during the model training process.

Expected Outcomes if the Project Fails:

We will still produce comprehensive datasets of Indonesian teen psychology.
We will provide a "failed-fast" case study on the limits of running SLMs offline on low-end hardware.