Conviction AI

An AI-powered workspace that helps investment analysts turn research into impactful insights, faster.

Comment

Inspiration

Equity research analysts spend hours manually processing SEC filings, earnings transcripts, and news to create structured research outputs. This workflow is fragmented across 30+ disconnected notes, emails, and models. Analysts waste hours rebuilding understanding every time they revisit a company—knowledge gets lost, context is rebuilt from scratch, and conviction weakens over time.

We built this application because we wanted to solve the real problem of fragmented research workflows that prevent analysts from focusing on high-value insights.

What it does

This is a web application that transforms manual equity research workflows into an intelligent, AI-powered workspace. Analysts search for companies by ticker and generate professional research outputs using Chrome AI's built-in knowledge, including:

  • Elevator Pitch: 30-second company summary for client presentations
  • Bull & Bear Case: Comprehensive investment analysis with structured arguments
  • Key Debates: Core investment debates and questions for decision-making
  • Meeting Prep: Pre-meeting briefing with recent developments and key metrics

For document analysis, users can upload earnings transcripts for AI-powered summarization and analysis. All outputs are formatted for professional use and can be exported to Google Docs, Notion, or Slack.

The core experience: Search for any company → Generate instant analyst-ready outputs using Chrome AI's native knowledge → Export professional analysis.

How we built it

We built this with Next.js 15, TypeScript, and Chrome's built-in AI APIs. The application uses:

  • Writer API: Generate structured investment analysis using Chrome AI's built-in knowledge of companies, markets, and financial concepts
  • Summarizer API: Process earnings transcripts and long documents into key insights with automatic chunking for large files
  • Proofreader API: Ensure all generated content meets professional standards
  • Rewriter API: Improve and refine content directly within the editor interface

Each analysis type (bull/bear cases, meeting prep, key debates) uses specialised prompts that leverage Chrome AI's native knowledge. For document analysis, the Summarizer API processes earnings transcripts, which are then analysed with structured prompts to generate professional outputs.

Technical architecture: We implemented intelligent document chunking to handle large files (>20KB), comprehensive error handling for API availability, and real-time progress tracking. All processing happens client-side with no external API calls—sensitive financial data never leaves the device.

Challenges we ran into

The main technical challenge was designing effective prompts for financial analysis and handling large documents. Earnings transcripts can be very long, so we implemented automatic chunking to process them in sections using the Summarizer API. We also had to ensure all processing happens locally without external API calls, which required careful error handling and availability checking.

Accomplishments that we're proud of

We achieved 100% local processing with Chrome AI APIs, generating professional analyst-ready outputs including bull/bear cases, key debates, and meeting prep in under 60 seconds. The application saves analysts 2-3 hours per company research cycle by transforming 30+ fragmented documents into a single structured workspace. We successfully implemented real-time progress tracking, automatic document chunking for large earnings transcripts, and professional output formatting suitable for client presentations.

What we learned

Designing for local AI requires rethinking how to leverage built-in knowledge effectively, we learned that Chrome AI's native knowledge of companies and markets is very comprehensive. Chrome AI APIs provide powerful on-device processing capabilities that enable new categories of privacy-first financial analysis tools.

What's next for Conviction AI

We plan to integrate backend storage for persistent analysis and connect to official financial datasets (SEC filings, earnings transcripts) to enhance Chrome AI's native knowledge. We're also exploring advanced model capabilities for more sophisticated financial analysis.

Built With

Share this project:

Updates