Devop

💡 The Problem
B2B sales prospecting and developer outreach require hours of manual overhead. Teams spend immense time jumping between tabs—scraping recent company announcements, analyzing product alignment, mapping target personas, and trying to draft pitches that don’t sound like generic AI spam.

🤖 The Solution
Auto-SDR completely automates this entire cognitive stack from a single input field. By utilizing Lyzr Architect, the system breaks away from the fragile “single-prompt blob” approach, instead decomposing the workflow into a robust, deterministic pipeline run by three specialized, autonomous agents.

🛠️ Architecture & Core Workflow
The application runs sequentially, passing structured states downstream between three independent agents:

The Investigator Agent (OSINT Researcher): Takes a single input (Company Name/URL) and uses autonomous web search and scraping modules to extract real-time context on the target’s recent technical pivots, hiring trends, or public objectives.

The Persona Mapper Agent (Strategic Analyst): Consumes the investigator’s structured brief and maps the company’s current pain points directly against an internal technical value proposition (e.g., a high-performance AI developer toolkit). It identifies the high-leverage buyer persona (e.g., VP of Engineering) and establishes the core psychological hook.

The Copy Architect Agent (Execution Writer): Synthesizes the analyst’s brief to craft a hyper-personalized, context-rich 3-part outreach sequence, delivering a flawless cold email, a targeted LinkedIn connection note, and a sharp 15-second elevator pitch in clean Markdown.

Tech Stack & Frameworks
Orchestration & Agents: Lyzr Architect (Multi-Agent State Management & Sequential Pipelines)

Frontend Architecture: Next.js, Tailwind CSS, TypeScript

Tool Calling: Autonomous Web Search & Web Scraping APIs

🚀 Why It Fits the Theme (Production-Grade Agentic Workflow)
True Autonomy: Moves entirely beyond basic chat prompting. The application executes a complex, long-horizon task with real-world tool execution and state transfers.

Engineered Craft: Relies on clear role decomposition and strict multi-agent handoffs, showcasing how modular AI agents can reliably replace a highly manual, multi-step knowledge workflow.