SupportIQ Chatbot
Instant answers for a nonprofit, built on their own knowledge.
SupportIQ Chatbot
Instant answers for a nonprofit, built on their own knowledge.
Project overview
During my web developer internship with Africa ICT Right, I helped redesign the website and built an AI chatbot trained on the organization's own content — programs, FAQs, contact paths — so visitors get answers instantly instead of waiting for email replies.
Problem
A small team was answering the same visitor questions repeatedly by email, while the website buried key information three clicks deep.
Idea
Pair a website redesign with a chatbot grounded in the org's real content — no hallucinations, always a path to a human.
Solution
A retrieval-grounded chatbot: questions are matched against curated org knowledge, answers cite their source pages, and anything uncertain escalates to the contact form with context attached.
My role
Web developer intern — redesign, chatbot build, knowledge curation, workflow improvements.
Main features
- Grounded answers from curated org content
- Source links on every answer
- Graceful handoff to human contact
- Website redesign for findability
- Internal workflow documentation
Challenges
Early answers drifted off-topic. Grounding every response in retrieved passages — and saying 'I don't know, here's who to ask' otherwise — fixed trust overnight.
What I learned
A chatbot is a UX project first and an AI project second. Escalation paths are a feature, not a failure.
Future improvements
- Multilingual support
- Analytics on unanswered questions
- Content-gap reports for the team