Client Overview
The client is an established player in the algo trading sector, running a long-term trading automation platform. Earlier development focused on the core trading logic and user dashboards.
The new Tickerbell AI initiative was designed to be a powerful, AI-driven strategy creation tool — capable of generating, refining, and adapting algorithmic trading strategies directly from user prompts. The goal was to make advanced algorithm writing accessible to all traders, even those without deep technical expertise.
The Challenge
The goal was to integrate this AI-powered assistant directly into the existing platform while keeping data security and system stability at the forefront.
Key requirements included:
- Secure, backend-level AI integration without exposing sensitive trading data
- Ability to pass unique user identifiers between frontend and AI logic for accurate, user-specific responses
- Reliable handling of real-time communication via WebSockets
- Context-aware strategy generation, even with incomplete parameter sets
- Backend-to-chatbot connection to replace the previous frontend-only integration
Our Solution
We delivered a backend-first AI integration built with Nest.js (backend) and Next.js (frontend), incorporating secure WebSocket communication and webhook validation.
Custom Features:
- User-ID based session mapping – every AI interaction is tied to a randomly generated user ID, ensuring data is routed to the correct session
- Secure webhook validation – environment-based secrets to verify data origins before processing
- Backend–AI communication layer – the AI assistant runs on the Voiceflow platform, which processes user prompts and sends structured responses to the backend for further handling
- Real-time updates – users connect to the WebSocket one layer earlier for faster response handling
Self-Management Toolkit:
We provided basic technical documentation and demo sessions so the client’s team could understand the integration logic and manage future adjustments independently.
Collaboration & Feedback
A compact two-person team — a project manager and a full-stack developer — worked closely with the client, holding weekly meetings and demos. This ensured rapid iteration and full visibility. The client appreciated the secure design and the stability of the integration, noting that it allowed AI-generated trading strategies to be created and modified in real time, without exposing core trading infrastructure.
Results
The project achieved its main goal: empowering users to write, adapt, and improve trading algorithms through a conversational AI interface.
Highlights include:
- Fully functional AI integration in under 3 months
- Stable, secure backend communication with AI logic
- Positive user reception, especially from less technical traders
- AI-assisted algorithm generation without direct database exposure
Technology Stack
- Backend & WebSocket: Nest.js
- Frontend: Next.js
- Integration Layer: Secure webhook with environment-based secrets
- AI: GPT-based conversational logic running on Voiceflow, integrated with the backend
Conclusion
The Tickerbell AI project demonstrates how a well-planned backend integration can enable advanced AI capabilities without compromising security or performance. By connecting a Voiceflow-powered AI assistant — capable of generating and refining algorithms — to the existing algo trading platform, the client can now empower traders of all skill levels to explore, create, and optimize strategies securely, and in real time.