StepToLife

An adaptive system for people in vulnerable situations that helps them find their way in a new environment and take things one step at a time

Comment

Inspiration

People in vulnerable situations — including individuals with cognitive or psychological differences, as well as those living below the poverty line, migrants, or minorities — often cannot use available opportunities because they don’t know how to act.

Key insight: People don’t lack opportunities — they lack clear steps to act.

What it does

StepToLife is an adaptive system that guides people in vulnerable situations through real-life challenges using simple, step-by-step actions.

Instead of overwhelming users with information, it first adapts to their current state — simplifying the interface, reducing cognitive load, and adjusting communication style based on how comfortable they feel.

Once a user selects a goal (like finding a job), the system generates a personalized action plan. It breaks the process into small, clear steps and supports the user in real time — helping them prepare, navigate, communicate (with phrases and translation), and complete actions.

We don’t provide information — we guide users to action.

How we built it

We built StepToLife as a lightweight web application focused on one clear goal: guiding users from confusion to real-world action.

Tech stack

  • Flask (Python) — backend logic, session handling, orchestration layer
  • JavaScript + HTML/CSS — fast, accessible, mobile-friendly UI

Core architecture

Core innovation: Orchestrated AI agents working as one system

  • Onboarding agent → understands user state and comfort level
  • Planner agent → builds a step-by-step action plan
  • Communication agent → generates phrases and supports interaction
  • Guidance agent → provides real-time instructions

How it works (end-to-end)

  1. The onboarding agent detects user needs
  2. The planner agent creates a structured path
  3. The communication agent prepares interaction (phrases, language)
  4. The guidance agent supports real-time actions

Challenges we ran into

The biggest challenges were not technical — they were about designing for real human barriers.

1. Adapting without asking sensitive questions

We needed to understand the user’s state without directly asking about mental health.

We couldn’t ask:

  • “Do you have anxiety?”
  • “Do you have a disorder?”

Instead, we used comfort-based and behavioral questions to adapt the experience.

The system dynamically adjusts:

  • color theme (calm vs standard)
  • font size and readability
  • amount of content on screen

This reduces cognitive load not only through content, but through the interface itself.

2. Orchestrating multiple AI agents

Instead of a single assistant, we used multiple agents.

The challenge was:

  • keeping shared context
  • avoiding contradictions
  • creating one continuous experience

3. Simulating real-world support

Our concept relies on real-time guidance:

  • navigation
  • communication support
  • translation

We had to simulate this in a simple but believable way.

4. Language and communication barriers

To support users like refugees, we needed:

  • translation
  • simple phrases
  • clear and usable output

Biggest challenge: Reducing real-world barriers without increasing complexity.

Accomplishments that we're proud of

Impact: Turning passive users into active participants.

StepToLife has strong and scalable user potential. It can support a wide range of users — from people with cognitive challenges to migrants and low-income individuals.

Value for the user

The user:

  • understands what to do
  • handles real-life situations
  • moves from problem to solution

Result:

  • finds a job
  • completes documents
  • accesses help or housing

Value for the ecosystem

StepToLife connects people who were not acting with organizations that can help them.

Employers

  • gain new candidates
  • reach people who previously didn’t apply

Educational centers

  • attract new participants
  • engage hesitant users

Government and services

  • receive real interactions
  • gain better insights into needs

NGOs

  • see real visits instead of passive traffic
  • use resources more efficiently

What’s next for StepToLife

1. Real-world integration

  • local NGOs and support centers
  • job and employment services
  • legal and social assistance

2. Smarter personalization

  • better UI adaptation based on behavior
  • improved step-by-step planning
  • learning from user actions

3. Partnerships

  • employers
  • educational institutions
  • NGOs and public services

4. Measuring impact

  • tracking completed actions
  • measuring real outcomes (jobs, services, support)

Final takeaway: StepToLife doesn’t just explain what to do — it helps people actually do it.

Built With

Share this project:

Updates