Sela Virtual Friend: Grant-Funded Psychological Support Chatbot for Teens by BotLabs Agency

Sela Virtual Friend — Confidential Psychological Support for Teens via Telegram

Project: Sela Virtual Friend Chatbot

Client: Ivan Franko International Foundation

Industry: Non-profit, Education, Mental Health, Grant-funded Projects

Sela Virtual Friend is a grant-funded Telegram chatbot designed to deliver confidential, interactive psychological support and educational content to teenagers, leveraging the personas of Ivan Franko and Lesya Ukrainka. Built for the Ivan Franko International Foundation, this AI-powered solution combines flexible content management, gamified interfaces, and strict privacy standards to address the mental well-being of young users in Ukraine and beyond.

---

The Roadblocks to Teen Psychological Support Before Sela

Traditional approaches to youth psychological support in Ukraine have faced significant barriers:

• Limited Accessibility: Many teens lack regular access to in-person counseling, especially in rural or under-resourced areas.
• Stigma & Confidentiality Concerns: Young people often hesitate to seek help due to fear of judgment or exposure of sensitive issues.
• Linear, Inflexible Content: School programs and static helplines rarely personalize responses or adapt to the user's mood, age, or gender.
• Resource Intensive: Human-led support lines require ongoing staffing and training, making it hard to scale outreach, especially during crises.

These factors resulted in teens either not reaching out at all, or dropping off after a single, unhelpful interaction. The Ivan Franko Foundation sought a solution that would feel approachable, relevant, and safe for Ukraine’s youth.

---

The Brief: Building an Empathetic, Flexible Virtual Assistant for Teens

BotLabs Agency was tasked with delivering a turnkey chatbot system that could:

1. Provide psychological support for teenagers via a friendly, interactive Telegram bot, using characters familiar from Ukrainian literature (Ivan Franko and, in the second phase, Lesya Ukrainka).
2. Enable dynamic, branching dialogue trees—not static scripts—to allow the bot to respond to user input, mood, gender, and age.
3. Deliver psychologist-approved, multimedia content, including videos acted by professionals in character, as well as stickers, text, and images.
4. Ensure strict confidentiality, with no storage of sensitive user questions, and options for automatic or on-demand chat history deletion.
5. Empower non-technical admins at the Foundation to add, edit, and sequence new topics and dialogue fragments independently, via a secure web admin panel.
6. Offer direct access to national hotlines and crisis resources via a simple UI element.

The solution needed to be:

• Grant-compliant and ready for future expansion
• User-centric with playful, accessible design for the 10–17 age group
• Easily scalable and maintainable by the Foundation’s team

---

---

The Technical Blueprint: Crafting the Sela Chatbot Ecosystem

BotLabs’ approach combined a modern conversational architecture with robust content management. The system was built and launched in two major phases:

• Phase 1: Development and release of core functionality with Ivan Franko as the first virtual character. Included content upload and logic for tree-based dialogues.
• Phase 2: Addition of Lesya Ukrainka as a second persona, expansion to 17+ major topics, and scaling to 700+ dialogue fragments.

Key elements of the solution:

• Telegram Bot: Messaging interface with buttons, text, stickers, images, and video support
• Personalization: User provides name, gender, and age, with content tailored accordingly
• Dialogue Trees: Complex scenario management via interconnected “fragments” (dialogue nodes)
• Admin Panel: Secure web app for content management and analytics
• Privacy Engine: No storage of sensitive conversation data; automatic and manual chat deletion
• Hotline Integration: One-tap access to national support services

---

Where Things Got Complicated: Overcoming Challenges

1. Designing for Teen Engagement & Trust

Challenge: Teens are notoriously difficult to engage, especially around sensitive topics. The bot needed to be both friendly and credible.

Solution:

• Used familiar literary characters (Ivan Franko, Lesya Ukrainka) voiced by professional actors in videos.
• Incorporated stickers, playful animations, and casual language.
• Personalized greetings and responses based on age/gender.
• Content was reviewed by certified psychologists for tone and appropriateness.

2. Building a Truly Flexible Dialogue Engine

Challenge: Admins required the ability to add and update topics, questions, and response branches independently—without developer intervention.

Solution:

• Created a modular "fragment" system: each dialogue node is an editable entity with configurable messages, media, buttons, triggers, and links to other fragments.
• The admin panel UI allows drag-and-drop reordering, branching, and real-time preview.
• Support for hundreds of interconnected fragments, enabling deep personalization.

3. Multimedia Content Delivery Within Telegram

Challenge: Baked-in Telegram limitations (e.g., file size, supported formats) and the need for fast, seamless delivery of video, images, and stickers.

Solution:

• Optimized all media assets for Telegram’s platform (compression, format conversion).
• Used CDN-backed hosting for instant video playback.
• Developed fallback text content for cases where media fails to load.

4. Maintaining Privacy by Design

Challenge: The chatbot must not store sensitive user conversation data—yet still provide analytics and allow for session management.

Solution:

• Implemented anonymous session tracking (no personal data stored).
• Allowed users to delete their entire chat history on demand.
• Automatic session expiry and chat clearance after inactivity.
• Admin analytics aggregate only non-personalized, depersonalized data.

5. Grant Compliance and Scalability

Challenge: The project required rapid onboarding, clear documentation, and the ability to expand (e.g., new personas, topics) without code changes.

Solution:

• Adopted Waterfall methodology with detailed scope and milestones.
• Provided training and onboarding for Foundation admins.
• Designed all components for modularity and future-proofing.

---

Building It Step by Step: From Idea to 700+ Dialogue Fragments

The Sela Virtual Friend chatbot was developed over two major phases, with core team roles including project manager, lead developer, content manager, UI/UX designer, QA engineer, and client-side content editors.

Project Timeline:

• Discovery & Planning: 2 weeks — requirements gathering, dialogue mapping, technical planning
• Development: 12 weeks — backend, Telegram bot, admin panel, content upload
• Content Integration: Ongoing — psychologists and actors prepared, reviewed, and uploaded materials
• Launch & Training: 2 weeks — user acceptance, admin training, go-live
• Expansion (Phase 2): 8 weeks — additional persona, more topics and fragments

Work Breakdown:

• Project Management: 100+ hours
• Development: 300+ hours
• Content Management: 50+ hours

Team Structure:

• 1 Project Manager (BotLabs)
• 2 Backend Developers (BotLabs)
• 1 Frontend/Admin Developer (BotLabs)
• 1 UI/UX Designer (BotLabs)
• 1 QA Engineer (BotLabs)
• 2+ Content Managers (Client/Foundation)
• Psychologists and Actors (Client/Foundation)

{{USER_PHOTOS}}

---

Results After Launch: Impact Beyond the Numbers

With the Sela Virtual Friend chatbot, the Ivan Franko Foundation achieved a scalable, interactive way to deliver psychological support and cultural education to Ukrainian teens:

• 700+ Dialogue Fragments: Covering 17 core topics, with each scenario tailored to user age, gender, and mood.
• 2 Literary Personas: Ivan Franko and Lesya Ukrainka, brought to life with professional video content.
• Confidentiality Guaranteed: All sensitive chat data is deleted automatically or on demand—ensuring trust and safety for young users.
• Multi-Modal Content: Videos, stickers, music, film recommendations, and 3D virtual museum tours.
• 24/7 Access: Teens can seek help any time, anywhere, directly from their phones.
• Direct Hotline Access: Immediate links to national psychological and safety hotlines.
• Admin Empowerment: Client team can now independently add, edit, and launch new topics as needed.
• Grant Compliance: Delivered on time and within scope, serving as a model for future grant-funded digital mental health initiatives.

---

---

Why This Was Only Possible With BotLabs: Lessons and Takeaways

BotLabs Agency’s unique approach made the Sela Virtual Friend project a success where others may have struggled:

• End-to-End Expertise: From requirements analysis to UX, backend infrastructure, and admin UI, all work was handled by a single, tightly coordinated team.
• True Partnership: Ongoing collaboration with psychologists, content creators, and the Foundation’s staff ensured the solution met real user needs—not just technical specs.
• Rapid, Adaptable Delivery: BotLabs delivered a full-featured chatbot and admin platform in under four months, with full transparency and detailed documentation for grant compliance.
• Future-Proof Design: The modular architecture means the Foundation can extend the bot with new topics, personas, and content—without developer bottlenecks.
• Security & Privacy by Design: Automated chat clearance and no data retention set a new standard for confidentiality in youth support chatbots.

Key Lessons:

• User-Centric Design Drives Engagement: Teens responded best to content that felt personal, playful, and culturally relevant.
• Admin Empowerment = Long-Term Success: Providing non-technical stakeholders with real control over content is crucial for sustainability.
• AI and Automation Are Transforming Social Impact: Grant-funded projects can now reach wider audiences with higher quality, real-time support—at lower cost.

The Sela Virtual Friend project demonstrates how thoughtful chatbot development—grounded in collaboration, privacy, and technical excellence—can deliver measurable value to non-profits and the communities they serve.

---