Blue Hex Software’s Post

For a rough chuckle ask your favorite chatbot (that has cross thread context): Tell me a joke using all the context you have gathered from my threads.

Imagine having an assistant that is available 24/7, and can answer all of your questions! Our partner Zavarovalnica Triglav, d.d. doesn't have to imagine anymore, because together we developed TISA, a unique chatbot designed to revolutionize their internal support. Take a look at what this chatbot can do!  

Are you hands-on building in GenAI and LLM space? How do you currently validate the model(s) output in different environments? A common approach is using a golden dataset but do you manually verify or wait for customer feedback? If you are in this space, I’d love to chat.

Take a look at what this chatbot can do!!!

Does your client-facing team have an AI chatbot helping them find answers and execute tasks based on past client conversations? Kaizan's users do! We released V1 on individual meetings, then V2 on all meetings that happened with a client, and V3 comes in Sept - All Calls, Emails, Chat messages and personalised recommendations specific to those stakeholders. Growth since feature launch 😁

Wondering how to get a chatbot to follow a series of step-by-step instructions? Check out this quick tutorial on how to write a lightweight prompt to achieve this! https://lnkd.in/gu5dizVK

GenAI is made for making weird and interesting concepts quickly & refining the outputs from there. You can make something quite cool after adjusting your models and parameters for a few hours. The best part is that variants can be created in just a few minutes once the Tools used: - ComfyUI - Topaz Adapted from Downbink workflows 🔊

💡 For production applications, it is crucial to evaluate the correctness of LLM responses that come from your own data. My colleague, Yves Vancraenendonck from MbarQ, wrote about evaluation frameworks for these kinds of apps that use RAG (retrieval augmented generation). Check it out below ⬇️ ⬇️ #llm #ai #rag #evaluation

Before putting your chatbot/RAG application into production, you want to be sure that it behaves as expected. In this article I describe how we at MbarQ do such evaluation using Large-Language Models.

Yves's words ring true: 'If you don’t measure it, you can’t improve it.' Trust is key, and that's where AI steps in – helping us deliver solutions our clients can rely on. An interesting read if you want to know more about how we do this! 👇