# GPT Synthesizer **Collaboratively implement an entire software project with the help of an AI.** GPT-Synthesizer walks you through the problem statement and explores the design space with you through a carefully moderated interview process. If you have no idea where to start and how to describe your software project, GPT Synthesizer can be your best friend. ## What makes GPT Synthesizer unique? The design philosophy of GPT Synthesizer is rooted in the core, and rather contrarian, belief that a single prompt is not enough to build a complete codebase for a complex software. This is mainly due to the fact that, even in the presence of powerful LLMs, there are still many crucial details in the design specification which cannot be effectively captured in a single prompt. Attempting to include every bit of detail in a single prompt, if not impossible, would cause losing efficiency of the LLM engine. Powered by [LangChain](https://python.langchain.com/docs/get_started/introduction), GPT Synthesizer captures the design specification, step by step, through an AI-directed dialogue that explores the design space with the user. GPT Synthesizer interprets the initial prompt as a high-level description of a programming task. Then, through a process, which we name “prompt synthesis”, GPT Synthesizer compiles the initial prompt into multiple program components that the user might need for implementation. This step essentially turns unknown unknowns into known unknowns, which can be very helpful for novice programmers who want to understand an overall flow of their desired implementation. Next, GPT Synthesizer and the user collaboratively find out the design details that will be used in the implementation of each program component. Different users might prefer different levels of interactivity depending on their unique skill set, their level of expertise, as well as the complexity of the task at hand. GPT Synthesizer distinguishes itself from other LLM-based code generation tools by finding the right balance between user participation and AI autonomy. ## Installation - `pip install gpt-synthesizer` - For development: - `git clone https://github.com/RoboCoachTechnologies/GPT-Synthesizer.git` - `cd gpt-synthesizer` - `pip install -e .` ## Usage GPT Sythesizer is easy to use. It provides you with an intuitive AI assistant in your command-line interface. See our [demo](https://www.youtube.com/watch?v=zFJDQOtIFGA) for an example of using GPT Synthesizer. GPT Synthesizer uses OpenAIs `gpt-3.5-turbo-16k` as the default LLM. - Setup your OpenAI API key: `export OPENAI_API_KEY=[your api key]` **Run**: - Start GPT Synthesizer by typing `gpt-synthesizer` in the terminal. - Briefly describe your programming task and the implementation language: - `Programming task: *I want to implement an edge detection method from live camera feed.*` - `Programming language: *python*` - GPT Synthesizer will analyze your task and suggest a set of components needed for the implementation. - You can add more components by listing them in quotation marks: `Components to be added: *Add component 1: what component 1 does, component 2: what component 2 does, and component 3: what component 3 does to the list of components.*` - You can remove any redundant component in a similar manner: `Components to be removed: *Remove component 1 and component 2 from the list of components.*` - After you are done with modifying the component list, GPT Synthsizer will start working on generating the code for your project.
