BizPilot AI

AI-native Business Operating System that automates customer support, CRM, invoicing, payments, marketing, sales, scheduling, and business analytics using autonomous AI agents.

Comment

Inspiration

Small businesses lose growth opportunities because operations are fragmented and manual: customer support, lead follow-ups, invoicing, payment reminders, marketing execution, and performance reporting are often disconnected across tools. Over 70% fail due to operational overwhelm. We built BizPilot AI to empower small businesses to run with the speed, efficiency, and intelligence of large enterprises without increasing headcount.

What it does

BizPilot AI is an AI-native Business Operating System powered by Google Gemini. It deploys six domain-specific autonomous AI agents to automate end-to-end small business operations:

  1. Customer Support Agent: Automatically answers queries, creates tickets, and escalates complex issues.
  2. Sales Agent: Qualifies inbound leads, schedules meetings, and performs automated follow-ups.
  3. Marketing Agent: Drafts social media posts, writes email marketing campaigns, and plans promotions.
  4. Finance Agent: Generates Stripe-ready invoices, tracks payments, and sends collection reminders.
  5. Operations Agent: Assigns team tasks, tracks progress, and compiles daily status updates.
  6. CEO Agent: Synthesizes cross-departmental data to generate executive business performance reports and proactive recommendations.

How we built it

  • Frontend: Next.js, TypeScript, Tailwind CSS, and Radix UI / shadcn/ui.
  • Backend: FastAPI Python service architecture with containerized deployment on Google Cloud Run.
  • Database & Auth: Cloud Firestore, Firebase Authentication, and Firebase Storage.
  • AI Engine: Google Gemini API (gemini-1.5-flash) via the official Python google-generativeai SDK, utilizing system prompts and strict JSON schema parsing with confidence-aware decision thresholds.
  • Integrations: Stripe for payment processing and Resend for transactional email notifications.

Challenges we ran into

  • Ensuring multi-agent context coherence and preventing circular agent loops.
  • Designing a Confidence-Aware Decision Framework that distinguishes low-risk autonomous tasks (e.g., FAQ responses, draft invoices) from high-risk tasks requiring human-in-the-loop approval.
  • Enforcing strict JSON schema responses from LLM function calls with graceful fallback error handling.

Accomplishments that we're proud of

  • Seamlessly unifying 6 distinct autonomous agents into a single coherent business operating system dashboard.
  • Achieving sub-second response times for agent workflow orchestration.
  • Zero-cost operational serverless cloud architecture on Google Cloud Run & Firebase.

What we learned

  • SMB owners don't want standalone generic chat interfaces; they need structured, actionable automation that performs work directly in their existing business context.
  • Fine-grained confidence scoring is vital for establishing trust with AI-native operational software.

What's next for BizPilot AI

  • Expanding multi-channel agent triggers (WhatsApp Business API, Twilio SMS, direct email parsing).
  • Introducing usage-based AI token expansion packs and automated tax calculation integrations.

Built With

Share this project:

Updates