Building the Grammarly for AI Prompts.
We believe AI should do what you mean, not just what you say. Flux bridges the gap between human intent and machine understanding.
The Origin Story
The idea for Flux was generated from a very simple, frustrated thought: Why isn't there a Grammarly-type tool that can refine a prompt on the go?
We have tools that automatically fix our spelling, tools that correct our tone in emails, and tools that autocomplete our code. But when communicating with the most powerful intelligence engines ever created (LLMs), we are left staring at a blank text box, forced to guess the exact optimal syntax to get a decent output.
With that chain of thought, Flux was born.
The first stage was sculpting the architecture. Before writing any code, the exact deterministic pipeline needed to be defined—how could a system dynamically audit an input for missing parameters (like tone or audience) without ruining the user's workflow? After building the first iteration, conducting extensive user testing, and analyzing feedback over many iterations... Flux evolved into the robust engine it is today.
Seamless Workflow: The Chrome Extension
The ultimate vision was always about removing friction at the exact moment of creation. We have finally built and shipped the Flux Chrome Extension (currently under review in the Chrome Web Store).
The UX is exactly what the original "Grammarly for prompts" thought envisioned: Simply select the text in the input field of ChatGPT, Claude, Gemini, or anywhere else on the web. A small Refine with Flux button appears. Click it, and if you are logged in, the Flux engine takes over. It will quickly ask you any necessary clarifying questions, and then instantly generate a heavily-engineered, deterministic prompt that you can literally replace right into the input field with one click.
Built With AI, For AI
We don't just build tools for the AI ecosystem; we actively utilize them. Flux was conceptualized, designed, and coded end-to-end entirely using Artificial Intelligence. We believe that to build the best tools for interacting with LLMs, you must be a power-user of them.
Hitanshu Parekh
Hi, I'm Hitanshu. I'm a self-taught AI builder who tries to solve real-world friction using technology. By day, I work as a Product Management Intern at an insurance broking firm, and by night, I build zero-to-one software using AI.