-
-
Screenshot of the App
Inspiration
What it does🚀 Inspiration
The idea for Ask2Insight stemmed from a common pain point: data is everywhere, but accessing insights often requires technical expertise. We envisioned a world where anyone, regardless of language or background, could ask questions in their native tongue and instantly receive actionable, visual insights from large datasets. The rise of multilingual LLMs and agentic workflows inspired us to bridge that gap, turning natural language queries into autonomous, intelligent data workflows.
🧩 What We Built
Ask2Insight is a multilingual AI agent that transforms user questions into data-driven stories. The Gradio-based interface enables users to:
Ask questions in multiple languages. Receive answers in different tones (casual, executive, formal). Visualize insights with auto-generated charts. Export analysis to Excel reports.
The full pipeline includes:
Language-aware input processing. LLM-driven SQL generation for BigQuery. Dynamic chart selection and visualization. Tone-adaptive summarization of insights.
🧠 What We Learned
How to orchestrate LLM agents into a cohesive, multi-step pipeline—from query parsing to data visualization. The power and limitations of multilingual LLMs in real-world use cases. Handling ambiguity in natural language, especially across different cultures and languages. Building a user-centric Gradio UI that supports multiple tones and chart types while maintaining clarity.
⚔️ Challenges Faced
Ensuring accurate SQL generation across diverse natural language structures and languages. Handling chart recommendations that match the data type, intent, and user preferences. Dealing with multilingual edge cases, especially for languages like Chinese and Japanese, with unique syntax and formatting. Building a seamless UX while managing asynchronous workflows between the LLM, BigQuery API, chart generator, and Gradio frontend.
Built With
- ai
- gcp
- google-bigquery
- gradio
- llm
- python
Log in or sign up for Devpost to join the conversation.