aivah

A platform for creating interactive AI avatar agents for digital worlds, enhancing user engagement through multimodal interactions.

Aivah

Welcome to Iva. Learn how to use the platform. First, to start interaction with Avatar, click on the chat icon. It's located on the scene at top left corner. Type your message in the input bar. Talk to your avatar. Click on the headphone icon located at top left corner on the scene. It's the second icon. Now start the conversation. Click on call icon. Here options gear icon one access settings. Click the gear icon to change your Avatar's llm model agent brain or voice. Two manage conversations. View and manage your current and past conversations. Note that history varies with each agent brain. Three troubleshooting tip.

If your avatar isn't responding try switching the llm model or Voice or start a new conversation. Web XR mode if you're using an Oculus medaquest 3 or apple Vision Pro headset in the browser click this option to activate the immersive Mr experience. Access scenes click the scenes icon to change scenes and colors. Choose a pre-made scene that suits your needs. 

Avatar setup access it from the main navigation menu to create new avatars or customize existing characters by assigning names voice personas and background stories. Manage avatars view available and shared avatars or create new ones from this section.

Create new Avatar click Ready Player me setup to open the avatar Builder design and customize your character with hundreds of options or generate one by uploading an image. Give your avatar a name and choose its voice. Import existing Avatar if you already have a Ready Player me character simply copy its glb source URL here. 

Create Avatar click create to finalize your avatar. Agent brain access the agent base from the main navigation menu to create and customize your agent brain using multiple data sources and modalities. Create new agent upload documents or ADD web URLs to train your agent assign a unique.

Name to your agent brain. You can include multiple data sources and web URLs including YouTube video links. Upload multiple data sources and document types. Once all web URLs and data sources are added, click create to finish. All data source trained and created will show here. You can add more content and data sources to existing agent by updating its knowledge base. Click here on ADD content. 

Multimodal and vision agent. Activate multimodal capabilities by creating a multimodal agent brain. Toggle the switch to build a multimodal agent. Then upload videos, audios, images, text, and supported documents. Unlike

Traditional knowledge-based agents multimodal agents are only supported by Gemini 1.5 PRM and Gemini 1.5 pro flash MMA models. Companion Hub click the companion section in main navigation menu to start interacting with your avatars. After creating knowledge agents follow the interaction steps outlined earlier. You can interact with all your available avatars from here. 

Talk to llm inside options gear click talk to llm to directly engage with your chosen language models such as chat GPT or Gemini for advanced questions. Without using the knowledge agent you can interact with multiple llms within the same.

Conversation maintaining history and context. Web search agent click web search agent in the settings gear to enable smart conversations that fetch and summarize information from the web. It can work alongside talk to LLM by either bypassing the knowledge agent or integrating with it to intelligently prioritize searches. 

Immersive web browsing experience. Empower your avatar with web search agent similar to perplexity Google search or co-pilot. Enjoy an immersive browsing experience with 3D widgets as your avatar searches and summarizes web info. Click on source images to be redirected to original web pages.

Playground use playground to fine-tune and test scenarios creating the perfect persona for your agent. Interact with avatars customize voices and knowledge bases and choose from various llms to suit your needs. Tip on mobile access playground from the main navigation menu in the top left corner by selecting the playground icon. 

Share Avatar customize your AI Avatar agent behavior and interactions by providing Specific Instructions and prompts. Once satisfied publish your avatar with a single click. Access share avatar from the main navigation menu or the Avatar options drop down. 

Agent prompt Define your

Agents actions and personality with customized prompts. Effective prompting shapes its character by including interjections, restrictions, and grounding as needed. You can also set up guard rails here. 

Tip: Changes to the agent prompt automatically update across the entire system for that agent brain, both within the

Note, that there may be slight differences in text and video. Watch video for the original content.

Aivah Video