Inspiration

Every day, people spend hours on repetitive tasks like downloading attachments, organizing files, updating spreadsheets, and sending notifications, all at the same time. While automation platforms such as n8n are incredibly powerful, they still take a while to understand logic. We wanted to make automation as simple as having a conversation. This became Sayso, where users describe what they want in plain English, and AI generates the workflow for them.

What it does

Sayso is a multi-agent AI-powered workflow automation platform that transforms natural language into intelligent, executable workflows. Users simply describe what they want to automate, and Sayso generates a structured workflow that can be reviewed, edited, and executed. The platform supports integrations such as Gmail, Google Drive, Google Sheets, while providing a visual workflow interface and live execution monitoring. Instead of spending time manually building workflow logic, users can focus on their goals while Sayso handles the automation.

How we built it

We built Sayso using a Next.js frontend and a FastAPI backend. The frontend provides a visual interface for creating and monitoring workflows, while the backend is powered by a multi-agent AI pipeline. A Planner Agent converts natural language into a structured workflow specification, a Critic Agent reviews the generated plan, a Validator ensures the workflow is logically correct, and a Compiler converts it into an execution graph. Our asynchronous execution engine executes each workflow through reusable connectors, while a Healer Agent automatically recovers from supported failures. Authentication is handled through Firebase Authentication and Google OAuth, and Firestore stores workflows and history.

Challenges we ran into

One of our biggest challenges was making AI-generated workflows reliable. Large language models can occasionally generate inconsistent or invalid outputs, so we designed a validation pipeline that checks workflows before execution instead of blindly trusting AI responses. Another challenge we ran into was getting our google authentication for gmail, drive, and sheets to work. It worked sometimes, but other times it broke, so we spent a lot of time solving that issue.

Accomplishments that we're proud of

We're proud that we built a complete end-to-end workflow automation multi-agent AI platform in such a short amount of time. From natural language planning to workflow generation, execution, monitoring, and error recovery, every major component works together in a modular architecture.

What we learned

This project taught us that building reliable AI systems requires much more than calling an LLM. We learned how important validation, structured data models, modular software architecture, and execution safety are when integrating AI into real-world applications. We also gained valuable experience working with FastAPI, Next.js, Firebase Authentication, Google OAuth, React Flow, and multi-agent system design.

What's next for Sayso

Our next goal is to expand Sayso's connector library to support many more services beyond Gmail, Google Drive, and Google Sheets. We also plan to add conditional logic, loops, scheduling, additional AI-powered workflow suggestions, collaborative workflow editing, workflow templates, and support for custom APIs.

Built With

• asyncio
• css
• driveapi
• fastapi
• firebase
• firestore
• google
• google-gmail-oauth
• googleauth
• httpx
• networkx
• next.js
• oauth
• openrouter
• pydantic
• pytest
• python
• react
• reactflow
• tailwind
• typescript
• vercel