Robofy is built on a modern, serverless infrastructure designed for high-throughput messaging, enterprise-grade security, and low-latency AI reasoning.This page outlines how our system processes data, orchestrates AI responses, and scales under the hood.
Data Ingestion & RAG Pipeline#
To ensure the AI agent provides highly accurate answers based on your business data, Robofy uses a specialized Retrieval-Augmented Generation (RAG) pipeline featuring Hybrid Search.When you upload documents or URLs, the system extracts, chunks, and vectorizes the text into both semantic (dense) and keyword-based (sparse) vectors.Note: By utilizing Hybrid Search, Robofy ensures that agents understand both the broad semantic meaning of a customer's question, as well as exact product names, SKUs, or industry-specific acronyms.
Infrastructure & Core Stack#
Robofy’s backend utilizes a mix of containerized and serverless AWS architectures. This allows us to scale messaging throughput entirely separately from dashboard operations.Backend & APIs#
Core API: Built on .NET Core 8 (C#) for enterprise-grade stability, performance, and reliable handling of structured data workflows.
Serverless Scaling: AWS Lambda powers our high-scale messaging APIs and webhook ingestion, instantly spinning up to absorb traffic spikes.
Long-lived Workloads: AWS App Runner hosts the dashboard API, billing, and API key management.
Buffering: Amazon SQS acts as our backbone queue, smoothing out traffic spikes and ensuring downstream systems are never overloaded during high-volume messaging campaigns.
AI & Data Layer#
LLM Engine: Support for almost all LLM models via Vercel AI Gateway. Primary model is Google Gemini, providing state-of-the-art reasoning, code execution, and intelligent tool routing.
AI Observability: We integrate Langfuse to trace and monitor every LLM execution. This provides granular visibility to track token consumption, measure reasoning latency, debug complex tool-calling sequences step-by-step, and continuously evaluate the quality of AI responses.
Vector Database: Pinecone powers our Hybrid Search capabilities for lightning-fast knowledge retrieval.
AWS RDS (SQL Server): Handles relational data (accounts, billing, RBAC).
AWS DynamoDB: Manages high-velocity message logs and webhook events.
AWS S3: Provides secure object storage for file uploads, exports, and document persistence.
Channels & Frontend#
Dashboard & Web App: Built on Next.js and hosted on Vercel with Tailwind CSS for a fast, developer-focused interface.
Web Chat Widget: Powered by the high-performance Vercel AI SDK for seamless streaming responses.
WhatsApp: Fully verified, compliant integration using the Official Meta WhatsApp Cloud API.
Transactional Email: Amazon SES handles alerts, verifications, and system notifications.
Security, Privacy & Compliance#
We treat your business data and API access with the highest security standards.Authentication: Managed by AWS Cognito. Supports email/password, Multi-Factor Authentication (MFA), and OAuth. Robofy layers a custom multi-tenant organization model on top for role-based access control (RBAC).
Encryption: All files stored in AWS S3 are encrypted at rest using industry-standard AES-256 protocols. All data in transit is encrypted via TLS 1.2+.
Data Privacy: Your business data and conversation logs are strictly isolated. We do not use your private knowledge base documents or customer conversations to train public AI models.
Monitoring & Logging: Entirely monitored via AWS CloudWatch to track message delivery, API latency, and identify anomalies instantly.
Modified at 2026-05-02 10:01:37
