AgentForge AI

AgentForge AI is an AI Workforce Compiler. A business owner describes their business by typing—or speaking—the biggest repetitive task per department. AgentForge analyzes the business, optionally researches live market data, designs a multi-agent AI workforce, and generates a downloadable, deployable codebase that runs anywhere with the owner's API keys. It is not a chatbot or hosted agent platform—it compiles and exports real code, avoiding lock-in. Core Features: Business Wizard — A guided form captures industry, team size, departments, and repetitive tasks. Voice input is optional; typing always works independently. AI Business Analysis — Converts input into a structured IR: departments → processes → tasks, with each task tagged by automation potential. Live Market Research — Uses Bright Data to gather competitor and industry data before agents are designed. AI Workforce Design — Generates 4–6 agents with roles, responsibilities, tools, and explicit handoff relationships. Interactive Org Chart — Clickable agent cards are grouped by department and connected with handoff lines. Selecting an agent opens its details. Agent Details — Each agent receives an editable 200–400 word system prompt and tool list, generated once rather than used as a chat interface. Working Research Agent — The Market Research Agent is genuinely wired to Bright Data MCP and runs real research after export, unlike the other tool stubs. Code Generation — Creates a OpenAI Agents SDK project with Dockerfile, docker-compose.yml, requirements.txt, prompt files, tool stubs, orchestrator, handoffs, and README. One-Click Export — Packages the workforce as a downloadable ZIP named after the business. Resilient AI Gateway — Every AI call is schema-validated, auto-repaired when invalid, cached to reduce spend, and falls back from AI/ML API to OpenRouter on failure. Cost & Reliability — A live panel shows usage, latency, fallback events, and repair counts for every generated agent.