BugVoyant

Reduce engineering time and simplify compliance: BugVoyant uses AI to auto-generate Sentry error post-mortems and anchors them on Algorand blockchain for an immutable audit trail.

Comment

Inspiration

Our team, composed of engineers who've spent countless nights on-call, knows the pain of incident response firsthand. After a critical alert fires, the tedious process of digging through logs, understanding the impact, and writing a post-mortem begins. This manual work is not just slow; it's prone to error and lacks the verifiable proof required for modern compliance standards like SOC 2. We were inspired to build a tool that we desperately wished we had: an "SRE on-demand" that doesn't just notify you of a problem, but provides immediate, actionable intelligence and an immutable audit trail. We wanted to transform the reactive stress of incident management into a proactive, intelligent, and auditable process.

What it does

BugVoyant-Ledger is an automated incident intelligence platform that transforms raw Sentry errors into actionable insights with a cryptographically verifiable audit trail. When an incident occurs:

  1. AI-Powered Analysis: Our resilient, multi-provider AI gateway (using Gemini, GPT-4o-mini, and Claude Haiku) instantly generates a detailed post-mortem report in Markdown, complete with a root cause analysis and recommended action items.
  2. Blockchain Anchoring: A unique SHA-256 hash of the report is anchored on the Algorand blockchain via the Nodely API, creating a permanent, tamper-proof timestamp and proof of existence for compliance and auditing.
  3. Audio Summaries: An AI-generated, 90-second audio brief of the incident is created using ElevenLabs, perfect for on-the-go executives and on-call engineers.
  4. Instant Notification: A rich, detailed alert is sent to Slack containing the AI summary, a link to the blockchain proof, and the audio file.
  5. Secure Dashboard: Users can manage projects, review report histories, and access audit logs through a secure, multi-tenant dashboard built with a robust client-side encryption model for storing all sensitive API keys.

How we built it

BugVoyant-Ledger is a modern, full-stack application built entirely on the sponsor's stack, demonstrating a professional, production-grade architecture.

  • Frontend: The user-facing dashboard was built using the Bolt.new Vite/React template, providing a polished and responsive user experience. It handles all project configuration and data visualization.

  • Backend: The core logic lives in a single, powerful Netlify Function. This serverless function is responsible for handling the incoming Sentry webhook and orchestrating the entire incident processing pipeline. This approach directly satisfies the Deploy Challenge.

  • Database & Auth: We used Supabase as our complete backend-as-a-service. This includes:

    • Postgres DB: For storing all user, project, and report data, with a comprehensive set of versioned migrations.
    • Supabase Auth: For secure user authentication, including email/password, SSO (Google/GitHub), and a custom Algorand wallet sign-in flow using a Supabase Edge Function.
    • Supabase Storage: For securely storing the generated audio files from ElevenLabs.
    • This deep integration makes us a strong contender for the Startup Challenge.

  • Resilient AI Gateway: We built a custom "AI Gateway" that intelligently routes requests to our primary provider, Google's Gemini 2.5 Flash-Lite, and automatically fails over to OpenAI and Anthropic models. This entire system is wrapped in circuit breakers and exponential backoff retries for enterprise-grade reliability.

  • Blockchain Integration: We use the algosdk to anchor report hashes on the Algorand Testnet. All API calls are routed through the Nodely API for unlimited, high-performance access, fulfilling the Blockchain Challenge.

  • Audio Generation: The ElevenLabs API is used for text-to-speech conversion, making our project eligible for the Voice AI Challenge.

  • Security: We implemented a sophisticated client-side encryption service using the Web Crypto API to ensure user API keys are encrypted before being stored in Supabase, providing an extra layer of zero-trust security.

Challenges we ran into

Our biggest challenge was architecting a system that was not just functional, but also secure, resilient, and observable—all within the tight timeline of a hackathon.

  • Enterprise-Grade Resilience: Building the multi-provider AI gateway was complex. We had to implement our own circuit breaker and retry logic from scratch in our resilience.ts module to ensure the system could gracefully handle failures from any single AI provider.
  • Zero-Trust Security: Implementing true client-side encryption for user secrets was a significant undertaking. We had to design a system where our backend never sees a user's raw API keys, which required careful management of a client-side encryption key in sessionStorage and custom Postgres functions for server-side verification.
  • Web3 Authentication: Creating the "Sign in with Algorand Wallet" flow required building a custom Supabase Edge Function to verify cryptographic signatures and issue JWTs, which was a deep dive into both Web3 and Supabase's auth internals.

Accomplishments that we're proud of

We are incredibly proud of building what we believe is a truly production-ready application, not just a proof-of-concept.

  1. Shipping a Complete Product: We didn't just build a feature; we built a full platform with a marketing homepage, multi-modal authentication (SSO, Web3, email), a feature-rich dashboard, and a robust, asynchronous backend.
  2. The Resilient AI Gateway: Our failover system across Google, OpenAI, and Anthropic is a piece of engineering we're particularly proud of. It makes our service reliable enough for real-world use.
  3. Our Security Architecture: The client-side encryption model is a testament to our "security-first" mindset. We are proud to have built a system that protects user data at every layer.
  4. Deep Sponsor Integration: We successfully and meaningfully integrated technologies from Bolt.new, Netlify, Supabase, Algorand (via Nodely), and ElevenLabs, showcasing the power of the modern web stack.

What we learned

This hackathon was an incredible learning experience. We went deep on what it takes to build a modern, AI-powered SaaS application. Key learnings include:

  • The power of serverless architecture (Netlify Functions) for building scalable, event-driven systems.
  • The incredible productivity boost of using a BaaS like Supabase for handling auth, database, and storage.
  • The nuances of designing resilient systems that can withstand third-party API failures.
  • The practical application of blockchain technology for a real-world business problem (auditing), moving it beyond just speculation.
  • The importance of cost-optimization in AI, which led us to select Gemini Flash-Lite as our primary model.

What's next for BugVoyant

We genuinely believe BugVoyant-Ledger has the potential to become a real, venture-backed company. Our roadmap is focused on turning this winning prototype into a market-leading product.

  • Production Launch: Move from testnets to mainnets, complete our payment integration (Stripe/RevenueCat), and launch our tiered subscription plans.
  • Expanded Integrations: Add support for other error monitoring services like Datadog and Rollbar, and other notification platforms like Microsoft Teams.
  • Advanced Analytics: Build out the analytics dashboard with MTTR (Mean Time To Resolution) trends, error-type clustering, and cost-management features for users.
  • Compliance Automation: Generate exportable SOC 2 / ISO 27001 compliance reports that automatically include the blockchain proofs for all incident responses, creating a "must-have" tool for regulated industries.

Built With

Share this project:

Updates