AI Agents & Automation

Savantic AI Lab operates as a full-stack AI lab, combining deep scientific expertise with real-world application to develop scalable, sustainable, and transformative AI solutions. With over two decades of innovation, they focus on "Meaningful AI" to drive sustainable growth, measurable impact, and long-term value across various industries. Their services range from research to real-world implementation, helping organizations turn AI potential into business impact. Savantic emphasizes ethical and responsible AI, ensuring solutions prioritize sustainability and deliver tangible results. They work with diverse sectors including Retail & Logistics, Medtech & Life Sciences, Industry & Energy, and Public Transportation & Municipalities.

AgentLabs offers an open-source universal frontend solution for AI Agents, enabling developers to quickly deploy their AI agents to public users. The platform provides essential features such as an authentication portal for user management, a clean chat frontend interface for user interaction, and integrated analytics and payment functionalities. Developers can control their AI agents with a real-time bidirectional streaming SDK from their backend, available in Python and TypeScript. AgentLabs aims to simplify the deployment process for AI agents, allowing developers to concentrate on the core AI logic while it handles the user-facing aspects. It supports both cloud hosting and self-hosting via Docker Compose, with an active alpha release and ongoing development.

Agently-Daily-News-Collector is an open-source project designed to showcase an automated daily news collecting workflow. Powered by the Agently AI application development framework, this tool allows users to input a topic and automatically generate a multi-column news briefing. The workflow includes searching, shortlisting, browsing, summarizing, and assembling stories into a final report, which is saved as Markdown. It features structured output contracts for clearer interfaces, built-in search and browse tools, and environment-aware settings for easy model configuration. The project emphasizes a clean app/workflow/tools/prompts split, enabling true concurrency in processing columns and summaries through TriggerFlow for efficient news collection.

agent-protocol offers a common interface for interacting with AI agents, addressing the challenge of diverse agent implementations. It provides an API specification, defined in OpenAPI, that agents can expose, making them interoperable regardless of their underlying framework. This protocol includes essential routes for creating tasks and executing steps, along with additional routes for managing tasks, steps, and artifacts. By adopting agent-protocol, developers can more easily benchmark agents, integrate them into other systems, and build general devtools for development, deployment, and monitoring. The project also provides an SDK for simplified implementation and a client library for users to interact with agents, fostering a more unified and efficient AI agent ecosystem.

AI Huberman is an AI-powered tool designed to simplify and provide insights from the extensive Huberman Lab podcast content. Users can ask science and health-related questions and receive AI-driven answers derived directly from podcast transcripts. The platform also offers timestamped links to relevant YouTube videos, allowing users to delve deeper into specific topics. Additionally, it provides personalized episode recommendations based on user queries and interests. This tool is ideal for anyone looking to quickly access and understand complex neuroscience and health information discussed in the Huberman Lab podcast, presented in clear, jargon-free language.

AI Reply is an AI-driven service designed to elevate brand presence on Reddit by automating engagement in relevant discussions. The tool continuously monitors Reddit for keywords, generates AI-powered reply suggestions based on article content, and allows users to send these replies to articles. This automated approach ensures brands maintain a continuous presence, saving significant time on manual monitoring and community management. It aims to boost brand visibility and impact within the Reddit community by ensuring brands are part of key conversations around the clock, effectively bringing in new customers.

Agent-ui is a modern chat interface designed for interacting with AI agents, built using Next.js, Tailwind CSS, and TypeScript. It offers seamless integration with local and live AgentOS instances through the Agno platform. Key features include a clean chat interface with real-time streaming, support for visualizing agent tool calls and their results, and the ability to display agent reasoning steps when available. It also handles multi-modality content like images, video, and audio, and provides references used by the agent. The UI is customizable with Tailwind CSS, and it's built on a modern stack including shadcn/ui and Framer Motion. Users can easily connect to their AgentOS instances, configure endpoints, and set up authentication.

AgentCPM is an open-source infrastructure developed by THUNLP, Renmin University of China, ModelBest, and the OpenBMB community, designed for training and evaluating various LLM agents. It addresses challenges in real-world applications such as limited long-horizon capability, autonomy, and generalization. The platform features AgentCPM-Explore, a 4B parameter deep-search LLM agent that achieves state-of-the-art performance on long-horizon benchmarks, and AgentCPM-Report, an 8B parameter deep-research LLM agent built on MiniCPM4.1-8B, capable of generating comprehensive reports comparable to top commercial systems. AgentCPM provides end-to-end open-source code for training, inference, and evaluation, along with a unified tool sandbox environment (AgentDock) for collaborative multi-model and multi-tool setups.

AI DIAL Core is an open-source project designed to provide a unified API for various chat completion and embedding models, assistants, and applications. Built on Java 21 and Eclipse Vert.x, it offers a robust and scalable solution for integrating diverse AI functionalities. The tool supports HTTP proxy functionality and provides comprehensive configuration options for static and dynamic settings, identity providers, toolsets, security, and storage. Developers can deploy DIAL Core on Kubernetes using Helm charts, making it suitable for complex enterprise environments. Its modular design allows for flexible integration and management of AI resources, ensuring a consistent interface across different AI services.

Kyugo is an innovative productivity tool designed to transform how users interact with their calendars, promoting unparalleled focus and efficiency. It moves beyond outdated spreadsheet-like interfaces by offering a visual and intuitive approach to time management. Kyugo allows users to define their day on their own terms, adapting to real-life dynamics rather than rigid schedules. It integrates with existing Google Calendars, providing features like instant visualization of busy times, thumb-friendly mobile navigation, and one-tap event rescheduling. The tool aims to empower users to take control of their day, offering a unique perspective on time as a canvas rather than a burden.

AutoDL-Projects is an open-source, lightweight project offering automated deep learning algorithms implemented in PyTorch. It provides various neural architecture search (NAS) and hyper-parameter optimization (HPO) algorithms, making it suitable for beginners, engineers, and researchers. The project features simple library dependencies, a unified codebase for all algorithms, and active maintenance. Key capabilities include implementations of NAS algorithms like TAS, DARTS, GDAS, SETN, NAS-Bench-201, and NATS-Bench, as well as HPO-CG. It requires Python >= 3.6 and PyTorch >= 1.5.0, with options for knowledge distillation and pre-trained models.

Backseat AI is an innovative companion app designed for League of Legends players, offering real-time AI coaching and commentary. It features personalized insights and advice from popular creators such as Tyler1, Emiru, and Jankos, making the gaming experience more engaging. The tool provides essential pre-game tips, including optimal picks, bans, and builds, alongside dynamic in-game overlays and post-game insights. Backseat AI is optimized to have minimal impact on PC performance, ensuring a smooth gaming experience. It is Riot-approved and compliant with their Terms of Service, ensuring players won't face bans for using it. The platform also supports content creators by sharing subscription fees, fostering a unique collaboration between AI technology and the gaming community.

Brainstory is an AI-powered productivity tool designed to help users articulate and structure their thoughts into compelling narratives. It transforms raw ideas into organized content in seconds, leveraging intelligent AI to streamline the creative process. The platform offers various pricing tiers, including a free option, and emphasizes ease of use with features like unlimited sharing and feedback responses. Brainstory aims to improve communication and mental performance by providing a structured way to talk through ideas, making it suitable for individuals looking to refine their thinking and content creation.

Vendra is a manufacturing marketplace designed to streamline the process of sourcing custom parts from U.S. manufacturers. It connects buyers with verified suppliers for services like CNC machining, injection molding, 3D printing, and sheet metal fabrication. Users can submit their CAD files and drawings to receive instant feedback on design issues and cost drivers. The platform then matches them with vetted manufacturers equipped with the necessary tools and capacity. Vendra emphasizes direct collaboration between buyers and shops, aiming to reduce costs by up to 30% compared to other online platforms. It also prioritizes security and compliance, being ITAR Registered, NIST 800-171 Compliant, and utilizing AWS GovCloud Infrastructure with AES-256 encryption.

BEHAVIOR-1K is a comprehensive simulation benchmark designed to accelerate Embodied AI research. It offers a platform for training and evaluating AI agents on a vast array of 1,000 everyday household activities, including tasks like cleaning, cooking, and organizing. These activities are meticulously selected from real human time-use surveys and preference studies, ensuring a human-centered approach to agent development. The monolithic repository provides all necessary components and dependencies, with a modular installation script allowing users to install only the required elements. This makes BEHAVIOR-1K an invaluable resource for researchers and developers in the field of embodied AI.

BehaviorTree.CPP is a C++17 library offering a robust framework for developing Behavior Trees, a powerful AI paradigm. It stands out by making asynchronous actions a first-class citizen, enabling reactive behaviors and concurrent execution of multiple actions. Trees are defined using an XML-based Domain Specific Language, allowing for runtime loading and dynamic morphology. The library supports both static linking of custom TreeNodes and runtime loading as plugins. It features a type-safe dataflow mechanism between nodes and includes a comprehensive logging/profiling infrastructure for visualizing, recording, replaying, and analyzing state transitions. This makes it ideal for complex AI implementations in robotics, game development, and as an alternative to Finite State Machines.

ChatGPT Auto Refresh is a browser userscript designed to maintain active ChatGPT sessions, effectively bypassing common issues like 10-minute time limits in Temporary Chat mode, network errors, and frequent Cloudflare security checks. By sending background requests, the tool ensures that conversations remain fresh and uninterrupted, even when the ChatGPT tab is not actively in use. This lightweight solution is built with minimal code, focusing solely on session persistence to conserve memory. It significantly enhances productivity by removing the need for constant copy/paste and manual refreshes due to inactivity, allowing users to focus entirely on their conversations. The tool is compatible with various browsers through userscript managers like Tampermonkey and Violentmonkey.

PlayThis is a unique platform that combines gaming culture with prediction markets, allowing users to forecast outcomes related to video games. Users can predict game release dates, Metacritic scores, awards, sales milestones, and more. By making accurate predictions, players earn PXP (PlayThis Experience Points) and climb leaderboards, with opportunities to win prizes. Beyond predictions, PlayThis also serves as a comprehensive gaming identity platform, enabling users to import their game libraries from Steam, Xbox, and PlayStation, track playtime, rate games, write reviews, and create curated lists. The platform fosters a community where users can propose new markets, vote on proposals, and engage in monthly prediction contests, making it a hub for both competitive prediction and personalized game tracking.

Federato is the first AI-native insurance platform designed to manage the entire policy lifecycle, from submission to quote and beyond. It leverages agentic AI to enable smarter risk decisions, accelerate quoting processes, and ensure portfolio alignment. The platform helps insurance carriers, MGAs, MGA aggregators, and mutuals to build and launch products faster with Product Studio, manage billing and payments, and achieve real-time portfolio steering through Control Tower. Federato aims to consolidate disparate systems into a single, proven solution, reducing time to quote and increasing high-appetite premiums. It provides a single pane of glass view for workflow management and real-time visibility into appetite and goal progress.

Enough Cream is an innovative app designed to help users achieve their perfect cup of coffee every time. Utilizing advanced computer vision and color analysis, the app analyzes the hexadecimal color value of your coffee as you add cream. Users first set a target by snapping a photo of their ideal coffee shade. Then, in real-time, the app compares the live pour against this profile, alerting you the exact moment to stop. It goes beyond simple light or dark analysis, considering saturation, hue, and brightness, and even adapts to ambient lighting conditions for consistent results. This ensures your coffee matches your precise taste preference, eliminating guesswork and delivering a consistently enjoyable morning ritual.

Una by Polydom is an advanced AI host specifically designed for the hospitality industry, including Airbnb, short-term rentals, long-term rentals, and hotels. It operates 24/7, handling guest communications across multiple channels such as phone calls (inbound and outbound), live chat via website widgets, email auto-responses, and messengers like WhatsApp, Telegram, and Facebook. Una integrates seamlessly with existing Property Management Systems (PMS) and Channel Managers to manage bookings in real-time, including creating, modifying, and canceling reservations. Beyond communication and bookings, it also coordinates tasks like housekeeping and maintenance for staff, tracking their completion. This AI solution aims to significantly reduce operational costs by providing an AI employee at a fraction of the cost of human staff.

dataset-api is an Open Source toolkit specifically developed for the ApolloScape Open Dataset, a comprehensive resource for autonomous driving research. It supports innovations across perception, navigation, control, and simulation for autonomous vehicles. The toolkit includes functionalities for trajectory prediction, 3D Lidar object detection and tracking, scene parsing, lane segmentation, self-localization, 3D car instance understanding, stereo estimation, and video inpainting. Researchers and developers can utilize this API to access and process large-scale datasets, facilitating the development and evaluation of advanced autonomous driving algorithms. The project is hosted on GitHub, providing code, examples, and detailed documentation for each subfolder.

DeepRec is a high-performance deep learning framework specifically designed for recommendation models, built upon TensorFlow 1.15, Intel-TensorFlow, and NVIDIA-TensorFlow. Developed since 2016, it powers core businesses like Taobao Search and advertising, offering robust features for training and inference. The framework excels in super large-scale distributed training, supporting models with trillions of samples and over ten trillion parameters. It includes in-depth performance optimizations for both CPU and GPU platforms, featuring advanced embedding variables, asynchronous and synchronous distributed training frameworks, and various runtime and graph-level optimizations. DeepRec also provides capabilities for delta checkpoint loading, super-scale distributed serving, and online deep learning with low latency.

DirectML is a high-performance, hardware-accelerated DirectX 12 library designed for machine learning tasks. It offers GPU acceleration for common machine learning operations across a wide array of supported hardware and drivers, including all DirectX 12-capable GPUs from major vendors. While DirectML is currently in maintenance mode, it remains supported on previous Windows releases and continues to ship with future Windows versions, receiving security and compliance fixes. It is distributed as a system component of Windows 10 and is also available as a standalone redistributable package for applications requiring a fixed version or running on older Windows 10 versions. DirectML exposes a native C++ DirectX 12 API and integrates as a backend for frameworks like Windows ML, ONNX Runtime, PyTorch, and TensorFlow, making it suitable for high-performance, low-latency applications such as frameworks, games, and other real-time applications.