Botdocs are datasets to train machines. Docs are provided in multiple formats to support Large Language Models, Intent Classifiers, and NLU engines, with data focused on the most common customer interactions.
Hey PHriends,
Botdocs is a free AI newsletter and I give these way for free to people that subscribe and refer others, but since some people don't like newsletters and just want the goods, you can bypass the free newsletter and buy the goods:
https://getbotdocs.gumroad.com/
AI is only as good as your data, but data is still a barrier to experiencing value from the technology. SMBs don't have enough data and enterprise have too much data, along with privacy and security concerns to accessing and sharing data with external systems.
Botdocs are a collection of datasets to train LLMs, intent recognition models, and NLU engines. The goal is to help both small and large businesses accelerate their time-to-value from AI customer experiences by focussing on the most common interactions. You just edit responses to common queries and a range of AI systems will ingest a Botdoc in one of 3 formats:
CSV - Most conversational AI systems ingest this format, plus it's easy to edit the responses. Just load it up in Google Sheets, edit, then export as CSV again.
JSONL - Fine-tune OpenAI models like Curie, Babbage, Ada, and Davinci. You can get comparable performance at a fraction of the cost of using GPT3.5 or GPT4. Some developers would make a case for fine-tuning with older models over embedding.
Dialogflow ES - NLU from Google where you need well structured data to get any value out of the technology but unlike generative LLM's, automated responses are never generated. It's the exact same response every time, and if you can't understand intent with Dialogflow, you can fallback to GPT, or another generative model.
Interested in Botdocs?
Feedback?
It's all much appreciated!
Roundups
Product Hunt Launch Checklist
CloudFunnels AI