Hello answers technical questions with simple explanations and relevant code snippets from the web.
Hey YC, weβre Michael and Justin - excited to launch Hello Cognition today!
Software engineering runs on ad-hoc know-hows - the most valuable information is either in technical documentation or scattered around discussion boards and blogs. Google is too broad and shallow of a search tool to be great at suggesting actionable technical solutions. Understanding information about a complex technical problem still involves a lot of manual time and effort.
Hello is a search engine that extracts understanding + code examples from technical sources, bringing you information you can immediately act upon to make progress on the problem youβre working on.
For our honors theses at UT Austin, we researched prototypes of large generative language models that can answer complex questions by combining information from multiple sources. Training a sequence-to-sequence language model (T5 derivative) on our custom dataset designed for factual generation yielded much better results with less hallucination. After creating this prototype, we started actively developing Hello with the idea that searching should be just like talking to a smart friend.
When you submit a query, we pull and rerank raw site data from Bing, then extract understanding with our proprietary large language models. For extracting and ranking code snippets, we use BERT-based models. We finally use seq-to-seq transformer models to simplify all this input into a final explanation.
Start asking your technical questions at sayhello.so! Join our Discord or contact us at founders@sayhello.so for early access to new features + get involved in our product feedback and iteration process.
We're looking forward to hearing your ideas, feedback, comments, and what would be helpful for you when navigating technical problems!