This is a submission for the [Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine | https://dev.to/challenges/hacktoberfest-week4-2026-10-26
What I Built
HackWhisper — Ask. Speak. Explore Open Source. 🎙️
HackWhisper is an open-source, AI-powered inbound voice assistant built using Python, Gemma 4, Whisper, and Graphify during Hacktoberfest Hack Day Islamabad 2026.
Our idea was simple: What if developers could learn about Hacktoberfest just by making a phone call?
Instead of navigating documentation and searching through multiple FAQs, users can interact with HackWhisper through natural voice conversations.
The assistant uses Hacktoberfest information as its knowledge base, processes spoken questions, and generates relevant conversational responses.
For the Ghost in the Machine theme, we explored the idea of giving a voice to information that normally remains hidden inside software and documentation.
Graphify adds another dimension by allowing us to visualize the components and relationships behind our AI system.
Demo
HackWhisper demonstrates how open-source AI can power an interactive, voice-first knowledge assistant.
How it works:
- A user calls our existing inbound phone number.
- The Python voice agent receives the call.
- Whisper converts the user's speech into text.
- The system retrieves relevant information from the Hacktoberfest knowledge base.
- Gemma 4 processes the question and generates a contextual response.
- The voice pipeline converts the response into audio and delivers it to the caller.
- Graphify provides a visual representation of the project's architecture and component relationships.
Project Website: https://hackwhisper.tech
GitHub Repository: https://github.com/Urk-Khan/HackWhisper
Code
HackWhisper is an open-source project developed using Python and open AI technologies.
GitHub: https://github.com/Urk-Khan/HackWhisper
The repository is intended to help developers explore the implementation, understand the architecture, and extend the assistant for other documentation-based applications.
How I Built It
We started with our existing Python-based inbound voice agent infrastructure and developed HackWhisper around open-source AI components.
Technologies Used:
- Python: Backend development and voice agent integration.
- Gemma 4: Open-weight language model responsible for conversational intelligence.
- Whisper: Speech-to-text processing for incoming voice questions.
- Hacktoberfest Knowledge Base: Documentation and FAQs used to provide relevant information.
- Graphify: Interactive visualization of the project's architecture and component relationships.
- Existing Server Infrastructure: Hosting and running the AI agent.
Technical Workflow
The inbound phone system receives a call and forwards the conversation to our Python voice agent.
Whisper processes the incoming speech, allowing the system to understand the user's question.
The application uses the Hacktoberfest knowledge base to provide relevant context, while Gemma 4 generates a natural-language response.
The response is delivered to the caller through the agent's speech output pipeline.
We also used Graphify to visualize the architecture, making it easier to understand how the voice agent, AI model, and knowledge components interact.
Why Does Open Innovation Matter?
Open innovation was one of the main motivations behind HackWhisper.
By using open-weight and open-source AI technologies, we can explore, customize, and improve the underlying system instead of relying entirely on closed AI services.
Gemma 4 provides flexibility for experimenting with language-model configurations and deployment approaches.
Whisper enables speech recognition using an openly available model.
Graphify helps us communicate the architecture visually, making the system easier to understand and explore.
Open AI technologies also make it possible for other developers to adapt the same approach for different applications, including technical documentation, educational platforms, community support, and developer assistance.
We believe that AI should not only be useful but also understandable, customizable, and accessible to the developer community.
My Agent Session
During Hacktoberfest Hack Day Islamabad, we worked on integrating our Python voice infrastructure with open AI components.
Our development focused on connecting speech recognition, Gemma-powered conversational intelligence, Hacktoberfest knowledge retrieval, and Graphify visualization into a functional application.
One of our main learnings was how different open-source AI components can work together to create a practical voice-based user experience.
The Spirit of It
What makes HackWhisper interesting to us is the ability to turn static information into something people can interact with naturally.
Imagine calling an AI assistant and asking:
"Hey HackWhisper, how can I participate in Hacktoberfest?"
Instead of searching through multiple documentation pages, you receive a conversational response.
For us, this represents the spirit of Ghost in the Machine: transforming information hidden inside a system into an interactive voice experience.
And with Graphify, we can also make the structure behind that experience visible.
Prize Categories
Best Use of Gemma
HackWhisper uses Google's Gemma 4 open-weight language model as its core conversational intelligence component.
Gemma processes user questions and generates responses using contextual information supplied by the application.
Team & Acknowledgments
HackWhisper was developed collaboratively during Hacktoberfest Hack Day Islamabad 2026.
Team Members:
- Urk Khan — https://github.com/Urk-Khan | https://dev.to/urk_khan
- Faseeh ur Rehman — https://github.com/Faseeh112
Together, we built HackWhisper using Python, Gemma 4, Whisper, and Graphify to make open-source knowledge accessible through conversational AI.
Special thanks to the Hacktoberfest Islamabad organizers, MLH, and the open-source community for supporting this project.
Top comments (0)