ProposalAI

Background

Writing a project proposal from scratch is time-consuming — requirements gathering, scope definition, timeline estimation, and deliverable breakdown all have to come together into a coherent document. This prototype explores what happens when that entire process is handed to an AI agent: the user describes what they want to build, and the system produces a complete, structured project proposal without manual drafting.

How It Works

The user starts by describing their project — either by typing or speaking. ElevenLabs handles voice input, transcribes it, and feeds it into the pipeline. LangChain orchestrates the agent flow: it breaks the input into structured components (project scope, objectives, deliverables, timeline, estimated effort, tech recommendations), runs them through OpenAI, and assembles the outputs into a full proposal document. The result is a formatted, end-to-end proposal generated entirely from the initial description.

Architecture

LangChain for agent orchestration and prompt chaining — OpenAI for content generation across each proposal section — ElevenLabs for voice input — users can describe their project verbally — Next.js frontend for the user interface and proposal display

Key Features

Voice-driven project description via ElevenLabs — LangChain agent pipeline breaking input into structured proposal sections — Full proposal generation: scope, objectives, deliverables, timeline, tech stack recommendations — End-to-end output from a single user prompt — no manual editing required

Outcome

Demonstrated that a multi-step proposal workflow — normally requiring several hours of structured writing — could be collapsed into a single voice or text prompt with coherent, usable output. Built as a prototype to validate AI agent orchestration patterns with voice input integration.