PromptoDesk AI

Overview

PromptoDesk was an AI-powered platform that enabled organizations to create, manage, and deploy custom AI tools without requiring employees to interact directly with Large Language Models (LLMs) or write complex prompts.

The platform allowed teams to build reusable AI assistants, create prompt templates, integrate with external systems, and execute AI-driven workflows across different business functions.

The Problem

As organizations began adopting Generative AI, teams faced several challenges:

• Valuable prompts were scattered across documents, chats, and individual users.
• Employees needed prompt engineering knowledge to get consistent results.
• Organizations lacked a centralized way to create and manage internal AI tools.
• Business teams wanted AI-powered workflows without dealing with APIs or model configurations.
• Different use cases required different AI models, integrations, and execution flows.

The goal was to create a platform that allowed organizations to standardize and scale AI usage across teams.

The Solution

PromptoDesk provided a centralized platform where organizations could:

• Create and manage multiple AI-powered tools.
• Build prompt-based assistants using templates or custom prompts.
• Execute simple Q&A interactions as well as multi-step AI workflows.
• Connect with external services and integrations.
• Support multiple AI models for different use cases.
• Manage AI tools across multiple organizations through a multi-tenant architecture.

This allowed business users to leverage AI capabilities without needing direct access to underlying AI services.

Technical Challenges

AI Orchestration

Different AI tools required different prompt structures, configurations, and execution strategies. Building a flexible orchestration layer was essential to support multiple use cases without creating one-off implementations.

Multi-Tenant Architecture

The platform needed to support multiple organizations while keeping configurations, prompts, users, and AI tools isolated and secure.

Dynamic Prompt Management

Users could create custom tools and prompts, requiring the system to dynamically generate and execute AI requests while maintaining consistency and reliability.

Workflow Execution

Some use cases involved simple question-and-answer interactions, while others required multi-step workflows with integrations and business logic.

Multi-Model Support

The platform was designed to support different AI models depending on business requirements, requiring abstraction layers that prevented vendor lock-in.

Technologies Used

• Angular
• NestJS
• OpenAI APIs
• TypeScript
• REST APIs
• Multi-Tenant Architecture
• AI Prompt Engineering
• Workflow Orchestration

Outcome

PromptoDesk helped organizations move beyond ad-hoc AI experimentation by providing a structured environment for creating and managing AI-powered tools.

The platform enabled teams to standardize prompt usage, build reusable AI assistants, and integrate AI capabilities into everyday business workflows without requiring deep technical knowledge of Large Language Models.

I worked primarily on the backend platform and AI orchestration layer using NestJS while also contributing to selected Angular frontend features. My responsibilities included: * Integrating OpenAI models into the platform. * Designing APIs for prompt execution and AI tool management. * Building the backend orchestration layer responsible for processing AI requests. * Supporting multi-model AI interactions. * Implementing integrations and workflow execution capabilities. * Contributing to frontend features and user-facing functionality where required.