Built For Developers Who Want to Chat with Documentation

WhatsUpDoc is the solution to a frequently faced problem for hackers using ChatGPT or CoPilot - the AIs were trained on outdated documentation and Github Discussions. Software moves faster than new models can be trained, so we augmented the base GPT4-Turbo model with a vector database of the up to date documentation and tribe knowledge.

tldr; WhatsUpDoc uses RAG (Retrieval Augmented Generation) to reduce hallucinations, and guide the LLM to provide accurate, up-to-date answers for your project's tech stack.

For example, since GPT4 was trained with a knowledge cutoff in 2021, it has no idea about NextJS's apps router. Go ahead, ask ChatGPT about the new apps router.

The same can be said about all sorts of modules and libraries out there. You could make a custom GPT on ChatGPT but they only support 5 pages worth of documentation.

WhatsUpDoc indexes every single page of the latest documentation every month, and the latest year's worth of Github Discussions.

We are adding integrations all the time. All paying customers will get access to all new integrations as they happen - no need to pay more.

We of course prioritize requests for integrations from paying customers.

What is Retrieval-Augmented Generation? Retrieval-Augmented Generation (RAG) is the process of optimizing the output of a large language model, so it references an authoritative knowledge base outside of its training data sources before generating a response.

Known challenges of LLMs include:

• Presenting false information when it does not have the answer.
• Presenting out-of-date or generic information when the user expects a specific, current response.
• Creating a response from non-authoritative sources.
• Creating inaccurate responses due to terminology confusion, wherein different training sources use the same terminology to talk about different things.

You can think of the Large Language Model as an over-enthusiastic new employee who refuses to stay informed with current events but will always answer every question with absolute confidence. Unfortunately, such an attitude can negatively impact user trust and is not something you want your chatbots to emulate!

RAG is one approach to solving some of these challenges. It redirects the LLM to retrieve relevant information from authoritative, pre-determined knowledge sources. Organizations have greater control over the generated text output, and users gain insights into how the LLM generates the response.

Source: https://aws.amazon.com/what-is/retrieval-augmented-generation/#:~:text=Augmented%20Generation%20requirements%3F-,What%20is%20Retrieval%2DAugmented%20Generation%3F,sources%20before%20generating%20a%20response. 

ChatGPT is a general purpose chat assistant. It has a wide knowledgebase and therefore good general knowledge of everything, but it will make mistakes when prodded for very specific things. You also need to fiddle with the prompts and mileage may vary depending on what your tech stack is.

WhatsUpDoc does all the hard work of indexing custom data for your tech stack + optimizing prompts, long context reordering, redundancy filtering, etc. so that you get the most optimized, up-to-date code assistant for your specific tech stack.

1. CustomGPTs will only allow you to index up to 5 pages of data one time. WhatsUpDoc will recursively crawl every page of the documentation for each tech stack.
2. CustomGPTs only do basic retrieval on the custom data. WhatsUpDoc does some additional steps to make sure the retrieved data is useful, non-redundant, and up to date.
3. With CustomGPTs, you will have to have multiple chats open for each part of your tech stack. So if you're building an app with NextJS, Tailwind, Stripe, and Supabase, chances are you will have switch between 4 different chat tabs. WhatsUpDoc allows you to add the different knowledgebases to a single workspace and our AI's semantic router will decide for you which knowledgebase to retrieve from.
4. CustomGPTs will not cite its sources (or if it does, it won't be up to date). With WhatsUpDoc you can get the answer and also easily verify that its accuracy by clicking on the sources cited.