Changelog - v2.6.0
Introduced hybrid search, improved chat widget, and updated web scraper. New landing page and revised GPT-4 pricing structure.
Last edited: 12/25/2023We believe in creating a tool that speaks for itself, powered and shaped by your support and feedback. So please don't hesitate to reach out to us with your thoughts and suggestions.
āØ New
- Hybrid Search: So far, we have been using what's usually called "semantic search" to find the most relevant chunks of your documents for a given question. This method is very accurate overall, but it starts to fail in accuracy when the user is asking about a very specific topic or a term (imagine the user is asking about a specific term in your software guidelines). We now also support "keyword search", which is much more accurate in these cases. The best part is that you can use both search methods at the same time, i.e. "hybrid search". Hybrid search is default in all new chatbots but for your current chatbots you can enable it in their answer settings.
- New Landing Page: We've updated our landing page to better reflect our product and its features. Check it out at https://www.docuchat.io.
- GPT-4 Pricing: We've updated our pricing to reflect our new pricing for GPT-4 queries. Going forward, GPT-3.5 will consume 1 credit per message, while GPT-4 will consume 15 credits per message.
š ļø Improved
- Improved Chat Widget: We've made several improvements to the chat widget. It now loads faster and is more reliable.
š Fixed
- Fixed an issue that caused our analytics page to crash.
- Fixed an issue that prevented us from parsing certain PDF files.
- Fixed several bugs and issues.
Product Manifesto
Explore the purpose and promise of DocuChat. Here is an honest look at our vision, motivations, and aspirations for transforming information access.
Changelog - v2.6.1
Implemented reranking for improved answer accuracy. Added year selection option in analytics. Various bug fixes and improvements.