Ainsider #27

A lot of #AI releases:
RunywayGen3
Synthesia 2.0
New multimodal LLM Moshi

Latest powerful AI Tools added to Ainsider Tools Library

Runway & Ai Video Prompting Guide

Free Flowise Web Scraper AI Agent Template

Huge AI launches and releases from last week:

Runway Gen3 Model released

See the best & most interesting generations below:

Synthesia launched their 2.0 model

Synthesia launch their 2.0 model - the very advanced AI video avatars & communication platform:
- AI screen recorder with high interactivity
- Improved creation & translation workflows
- Next-gen AI avatars - soon with hands 👀

New very powerful multimodal LLM

Kyutai Labs has made a significant breakthrough with new LLM Moshi - a real-time multimodal AI model.

This open-source (soon) model can listen, speak, and understand emotions

Details:
- Listens and generates Audio/Speech
- Thinks as it speaks (textual thoughts)
- Supports 2 streams of audio to listen and speak at the same time
- Has a smaller variant that runs on a MacBook or consumer-size GPU.
- Used Joint pre-training on mix of text and audio
- Used synthetic data text data from Helium a 7B LLM from Kyutai
- Learned its voice from synthetic data generated by a separate TTS model
- Achieves a end-to-end latency of 200ms
- Uses watermarking to detect AI-generated audio (WIP)
- Will be released open source

Latest AI Tools added to Ainsider.tools AI Library

A new wave of user-friendly tools is empowering creators and developers of all skill levels to explore and leverage the power of AI.

Let’s delve into some innovative platforms that are making AI development more accessible:

REPLICATE

Replicate is a platform designed to make machine learning models accessible and easy to use for developers. It allows users to run machine learning models from their browser or via an API, providing an intuitive interface for integrating AI into various applications without requiring in-depth expertise in machine learning.

API Access: Easily integrate machine learning models into your applications using simple API calls.

Browser Interface: Run and test models directly from the browser without the need for extensive setup.

Model Library: Access a wide variety of pre-trained models for different tasks, including image generation, text processing, and more.

Collaboration Tools: Share models and results with team members to collaborate on AI projects efficiently.

Scalability: Handle varying workloads with scalable infrastructure.

URL: https://replicate.com

SCENE

Scene is an innovative AI website and apps builder. It is tailored for developers and designers who need to build immersive experiences for web and mobile applications, offering powerful tools for real-time rendering and collaboration.

Real-time Rendering: High-quality 3D rendering in real-time, ensuring smooth and responsive user experiences.

Collaboration: Multiple users can work together on the same scene, making it ideal for team projects.

Asset Management: Easily manage and organize 3D assets within the platform.

Integration: Compatible with various tools and platforms, facilitating seamless integration into existing workflows.

Customization: Extensive customization options to tailor the environment to specific needs and branding.

URL: https://scene.io

WORKATO

Workato is a leading automation platform that integrates applications and automates workflows across various systems. It is designed to help businesses streamline their operations by automating repetitive tasks and improving efficiency.

Features:

App Integrations: Connect with hundreds of applications, including CRM, ERP, and marketing tools.

Workflow Automation: Create complex workflows that automate tasks and processes across different applications.

Custom Triggers: Set up triggers to initiate workflows based on specific events or conditions.

Monitoring and Analytics: Gain insights into workflow performance with robust monitoring and analytics tools.

User-Friendly Interface: Intuitive interface that allows users to build and manage automations without coding.

URL: https://workato.com

GUMLOOP

Gumloop is a dynamic AI automation tool focused on business processes and needs, as creating for custom automations.

It offers businesses the ability to create personalized and interactive customer experiences across various workflows and processes.

Conversational AI: Build and deploy advanced chatbots that understand and respond to customer queries naturally.

Multi-Channel Support: Integrate with websites, social media platforms, and messaging apps to reach customers wherever they are.

Analytics: Track and analyze chatbot interactions to gain insights and improve performance.

Customization: Customize chatbot behavior and responses to align with your brand and business goals.

Seamless Integration: Easy integration with existing systems and workflows to enhance customer support and engagement.

URL: https://gumloop.com

SUPERWHISPER

Superwhisper is a state-of-the-art AI-driven transcription service that converts speech to text with high accuracy. It is ideal for professionals who need reliable transcription for meetings, interviews, and other audio content.

Features:

High Accuracy: Advanced algorithms ensure precise transcription of speech.

Multi-language Support: Supports multiple languages, making it versatile for global use.

Real-time Transcription: Provides real-time transcription for live events and meetings.

Editable Transcripts: Users can easily edit transcripts to correct any errors or customize the output.

Secure Storage: Ensures that audio files and transcripts are securely stored and accessible only to authorized users.

URL: https://superwhisper.com

GPT-4 ALL

GPT-4 All by Nomic.ai is an advanced language model built on GPT-4 technology. It offers powerful natural language processing capabilities for a wide range of applications, from content generation to complex data analysis.

Natural Language Understanding: Understands and processes complex language inputs to provide accurate and relevant responses.

Content Generation: Generates high-quality written content, including articles, reports, and creative writing.

Data Analysis: Analyzes large datasets to extract meaningful insights and trends.

Customization: Fine-tune the model to suit specific needs and applications.

URL: https://www.nomic.ai/gpt4all

RunwayML Gen3 Prompting Guide

Gen-3 Alpha has endless potential to bring your artistic vision to life - but creating a strong prompt that conveys the scene is the key to generating video aligned with your concept.

This article covers different example structures, keywords and prompting tips to help you get started with Gen-3 Alpha. These are just examples – don’t be afraid to experiment when bringing your ideas to life.

Prompt Structures

Base Prompt

Prompts are most effective when they follow a clear structure that divides details about the scene, subject and camera movement into separate sections.

Using the following structure should help provide consistent results while you’re familiarizing yourself with Gen-3 Alpha:

[camera movement]: [establishing scene]. [additional details].

Using this structure, your prompt for a woman standing in a tropical rainforest might look like this:

Low angle static shot: The camera is angled up at a woman wearing all orange as she stands in a tropical rainforest with colorful flora. The dramatic sky is overcast and gray.

Repeating or reinforcing key ideas in different sections of your prompt can help increase adherence in the output. For example, you might note that the camera quickly flies through the scenes in a hyperspeed shot.

Try to keep your prompt focused on what should be in the scene. For example, you would prompt for a clear sky rather than a sky with no clouds.

Sample Prompts

Seamless Transitions

Continuous hyperspeed FPV footage: The camera seamlessly flies through a glacial canyon to a dreamy cloudscape.

Camera Movement

A glowing ocean at night time with bioluminescent creatures under water. The camera starts with a macro close-up of a glowing jellyfish and then expands to reveal the entire ocean lit up with various glowing colors under a starry sky. Camera Movement: Begin with a macro shot of the jellyfish, then gently pull back and up to showcase the glowing ocean.

Text Title Cards

A title screen with dynamic movement. The scene starts at a colorful paint-covered wall. Suddenly, black paint pours on the wall to form the word "Runway". The dripping paint is detailed and textured, centered, superb cinematic lighting.

Prompt Keywords

Keywords can be beneficial to achieve specific styles in your output. Ensuring that keywords are cohesive with your overall prompt will make them more apparent in your output.

In example, including keywords about skin texture wouldn't be beneficial to a wide angle shot where the camera is not closely focused on a face. A wide angle shot might instead benefit from additional details about the environment.

While keeping this cohesiveness in mind, below are different keywords you can experiment with while drafting your prompts.

All details with camera types & movements: https://help.runwayml.com/hc/en-us/articles/30586818553107-Gen-3-Alpha-Prompting-Guide

Runway & Ai Video Prompting Guide

Download, use with Flowise - only pasting your urls in Web Scraper Node: