Create Chat Agents in Riku
We're all looking for paradise, right? With the evolution in AI and the rise of Chat UX, it is time to leverage some of these advancements within Riku and save you and your team a ton of time. You can now create chat agents with ChatGPT, GPT-4, and Claude and some of the things we have planned for the near future will make this process even more seamless.
What is a Chat Agent?
The new chat UX from OpenAI makes it easy for you to provide a system message. This is where a lot of the magic happens and you can go super deep into providing context and knowledge about the task that you want the AI to perform. Having a generic system message is similar to how we would talk about the older AI models in terms of generic input equals a generic output.
Think of it like this, if I want to use AI to answer emails I can put in a system message like the following;
You are an expert at answering emails for business, you answer all emails in a short and concise manner.
I've asked for a few specifics in that system message but I still haven't gone super deep on anything specific, compare the output to answering an email using that system message as to using one like the following;
You are an expert in answering emails for a founder called Stuart at the business called Riku.AI. Riku.AI is all about helping coders and no-coders make the most of all AI technology aggregating all of the best LLMs under one roof making it super simple to experiment, test and deploy directly in your business and life. You answer simply, concisely and try to help customers with their enquiries about AI whilst being super enthusiastic for the future.
The system message is where the magic happens and a lot of users don't realize that just yet so get ahead of the crowd by crafting beautiful system messages to create hyper-focused agents directly for your business.
How do I set up Chat Agents in Riku?
From the Riku dashboard, click Chat Apps in the sidebar menu. You can then hit the + button to be taken to the Chat Studio or click this link directly. Once you are here, you can start to build out your chat agent and app. Customize settings like the greeting message and placeholder to make it work for your brand and then choose which technology you want to use.
Chat Apps from Riku currently offer ChatGPT (GPT-3.5 Turbo), GPT-4 8K, and GPT-4 32K from OpenAI. We're also offering Anthropic's Claude model for users at no additional cost to celebrate the launch of Claude on Riku. When you have selected your technology, if it is an OpenAI model, you can determine your system message or choose from a few templates, and then you can save your Chat App. That was super simple.
Note that Claude is a more general-purpose model currently and doesn't have a system message to set up. It can perform incredibly well on many different writing and logic tasks and we encourage you to experiment with it.
Go forth and explore the Chat Agent you have created now by typing into the chat and seeing the response. You can add more sessions by clicking the button at the top of the chat and we even have ways for you to publish your chat apps for public use. Choose to make them public, embed them on existing pages or password protect to control who has access. We want Chat App building to be simple, accessible, and useful for businesses of all sizes.
Looking to the future
Creating chat agents is just the start of our plans in this area. We want your chat agents to be able to access and analyze your documents, know your business and be incredibly smart and personalized to your needs. With Riku's proprietary AI process, we'll enable a future where regardless of the chat question or interaction that occurs, the correct chat agent will be tasked with answering it.
If I were a hotel, I want to have a customer support chat agent talk with guests whereas I would want a different F&B agent to deal with room service requests. One interface, multiple possibilities. This is how you transform business and give yourself an unfair advantage.
Try it out for yourself today at Riku.AI, a place to experiment, test and deploy the best large language models all in a single place made simple for both coders and no-coders alike.