Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.promptjuggler.com/llms.txt

Use this file to discover all available pages before exploring further.

PromptJuggler is a platform for authoring prompts, building workflows and agents, evaluating quality, and running everything at scale. One place for the whole loop: write it, wire it, test it, ship it. You bring your own LLM API keys (OpenAI, Anthropic, Google Gemini). PromptJuggler handles the rest: versioning, orchestration, parallel execution, tools, knowledge bases, memory, and a clean API to plug it all into your product. Complex underneath, simple on the outside.

​
Why it exists

AI tooling today can feel like receiving a thousand puzzle pieces with no picture on the box. Evaluation means squinting at outputs and hoping for the best. Switching from one model to another means rethinking half your setup. And the moment you want to combine multiple steps into something coherent, the glue work multiplies faster than the actual value you’re trying to create. PromptJuggler is the organised workbench for people who want to build with AI properly: with care, with structure, and with the satisfying click of things fitting together exactly the way they should.

​
What you can build

Start simple. A single prompt with a template, model selection, and parameter tuning. Run it, refine it, version it. Try it on GPT, then flip to Claude or Gemini with one click. No SDK juggling, no redeployment. Then go further. Wire prompts into workflows: visual pipelines where data flows through transformation nodes, branches, loops, and parallel execution. Each piece does one thing well; together they can orchestrate remarkably sophisticated agents. Think LEGO for AI: simple bricks, serious structures, and the quiet satisfaction of watching them work.
Workflow canvas
Add knowledge. Upload documents to a knowledge base, and your prompts can search them at runtime. That’s RAG: no vector database setup, no embedding pipeline to maintain, no infrastructure headaches. Then prove it works. Evaluations let you run a prompt against a dataset and measure the results. When you change something, you’ll know whether it got better or worse. Stop guessing. And ship it. The REST API lets you trigger prompt and workflow runs, read results (status, output, cost, token usage), manage knowledge base documents, and receive webhooks when things finish. The execution engine is built for throughput: a thousand parallel prompt runs without breaking a sweat, on a fraction of the resources you’d expect. Build it in the UI, call it from your product, and let the platform do the heavy lifting.

​
Who it’s for

Anyone with a spark of curiosity for building systems. Whether you’re an engineer wiring AI into a product, a team lead prototyping that chatbot idea that keeps collecting dust on the backlog, or someone who simply wants an organised way to tame the chaos of working with LLMs: if you care about doing this well, you’re in the right place.

​
Next steps

Quickstart

From zero to your first prompt run in five minutes.

Settings

Add your LLM provider keys and get ready to go.

Core Concepts

Understand how prompts, workflows, and versioning fit together.

Integration Guide

Connect PromptJuggler to your own product via the API.