Sennbot: WhatsApp Virtual Assistant
Sennbot is an intelligent, high-throughput WhatsApp Virtual Assistant engineered to transform standard messaging loops into an automated workspace. Built on top of the powerful WhatsApp-Web.js (wawebjs) ecosystem, Sennbot breaks the limits of traditional chat utilities by injecting advanced Large Language Models (LLMs) directly into private chats and group streams.
The virtual assistant is engineered not just for basic text conversations, but as a multi-modal productivity engine. Users can seamlessly upload images, documents, and spreadsheets, engaging in context-aware follow-up discussions with the AI. Backed by an administrative dashboard system featuring granular user registrations and strict tokenized API quotas, Sennbot serves as a highly scalable, secure automation tool for both team collaborations and public broadcasts.
Role
Backend Developer
Duration
Challenge
Building a stable, context-rich, and multi-modal AI layer natively within the constraints of WhatsApp introduced several architectural challenges:
Session & Payload Stability: Maintaining continuous, low-latency WebSocket connections via client-side abstractions while handling massive media buffers (images, stickers, and document uploads) concurrently.
Contextual State Persistence: Engineering a stateful chat history tracker capable of keeping conversational context alive across sequential commands (e.g., maintaining reference data during deep file or image interrogations).
Quota & Security Enforcement: Protecting the underlying LLM from API abuse by building an internal, real-time transactional ledger to regulate dynamic user balances and track token usage directly through chat commands.
Solution
Multi-Modal AI Engine: Allows users to query the bot for general intelligence (
/ask), parse complex imagery (/image), process data sheets (/file), and ask deep context-aware follow-up questions (/imgres,/fileres).Automated Broadcast & Utility Hub: Streamlines high-impact administrative workflows with automatic multi-group image broadcasts (
/bc) and seamless image-to-sticker conversions (/sticker).Enterprise Operations & Group Management: Empowers administrators to programmatically generate groups (
/creategroup) and bulk-add targeted contacts using high-speed CSV file processing (/invitecsv).
Technical Engineering
Detailed features engineered for this project.
- Decoupled Messaging Infrastructure: Engineered an asynchronous automation pipeline utilizing WhatsApp-Web.js, establishing stable event listeners to process real-time media buffers and custom routing rules.
- Multi-Modal LLM Pipeline Integration: Developed advanced content-handling endpoints that ingest, clean, and convert incoming binary streams (PDFs, CSVs, Images) into structured token inputs for real-time generative responses.
- Stateful Context Memory Architecture: Implemented localized conversational state management, allowing the bot to retain specific file and media contexts across back-to-back command cycles (/imgres, /fileres).
- Tokenized Quota Ledger & Access Control: Architected a robust user-management system featuring manual self-registration (/self-regis), SuperAdmin overrides (/register, /addquota), and isolated real-time database queries to enforce strict request limits (/kuota).
Let's talk about your project!
