AI Agents & Automation

SimpleMem is an advanced memory framework designed for LLM agents, offering efficient lifelong memory capabilities for both text and multimodal data. It employs semantic lossless compression to store, compress, and retrieve long-term memories, ensuring high information density and token utilization. The system features a three-stage pipeline: Semantic Structured Compression, Online Semantic Synthesis, and Intent-Aware Retrieval Planning. Omni-SimpleMem extends these capabilities to include image, audio, and video, achieving state-of-the-art performance on benchmarks like LoCoMo and Mem-Gallery. It supports cross-session memory, allowing agents to recall context and learnings across conversations, and integrates with platforms like Claude, Cursor, and LM Studio via MCP or Python.

ASRT_SpeechRecognition is a comprehensive, open-source Chinese speech recognition system built on deep learning principles, utilizing TensorFlow.keras. It integrates deep convolutional neural networks, long short-term memory networks, attention mechanisms, and CTC for high accuracy. The system supports both training custom models and performing inference, with pre-trained models available for immediate use. It offers API servers for HTTP and gRPC protocols, enabling flexible integration into various applications. Developers can leverage its SDKs for different platforms and programming languages, including Windows, Python3, Golang, and Java, or use its RESTful Open API. ASRT is designed for technical users and developers looking to implement robust Chinese speech-to-text functionalities.

Navio Auto is at the forefront of developing advanced AI technologies for autonomous driving, aiming to revolutionize urban mobility and cargo transportation. The company's core mission is to enhance safety and efficiency through its innovative solutions. Key offerings include the PlayAuto software platform, which enables seamless interaction with autonomous vehicles both directly and remotely, and the development of L5 autonomous trucks for long-haul logistics. Navio Auto's technology undergoes rigorous testing to ensure reliable and safe operation, even in complex urban environments. The company actively collaborates with partners in the automotive and IT industries to integrate its autonomous systems, striving to make human life more comfortable and secure through future mobility solutions.

QANDY is an AI-native autonomous execution engine designed for cryptocurrency and digital asset trading. It operates 24/7, leveraging a Level 3 AI agent and a state-of-the-art trading model to manage digital assets like BTC, ETH, and SOL. The platform offers real-time AI decision-making, adaptive routing across both decentralized (DEX) and centralized (CEX) exchanges, and risk-sensitive portfolio optimization. Built with institutional-grade architecture, QANDY aims to provide superior crypto AI trading compared to traditional platforms by autonomously analyzing market conditions and executing trades with precision, minimizing slippage and maximizing liquidity opportunities.

xiaozhi-android-client is an open-source, multi-platform AI assistant application built with Flutter, designed for Android, iOS, Web, Windows, macOS, and Linux. It enables real-time voice and text interactions, supporting various AI service providers like OpenAI and MiniMax. Key features include adaptive themes, an innovative mood mode, real-time voice interruption, and the ability to manage multiple AI assistants. The application also offers advanced functionalities such as OpenAI interface speed testing, Live2D model switching, and IoT integration for controlling phone functions. It provides comprehensive user management, including device, role, and voiceprint management, along with conversation and memory records. The commercial version integrates deeply with a self-developed backend, offering features like sub-second response times, MQTT protocol support, voice cloning, and voiceprint recognition, making it a robust solution for personalized AI assistant development.

Emesent is a pioneer in Physical AI, combining artificial intelligence, autonomous robotics, and advanced spatial computing to enable machines to understand, navigate, and map complex real-world environments without human intervention. Their solutions, including the GX1 and Hovermap, offer fast and accurate mobile mapping for hazardous and GPS-denied environments. Emesent's technology provides detailed mapping insights, advanced autonomous mapping capabilities, and versatile plug-and-play accessories for diverse mapping missions. It serves various industries such as mining, AEC, defense, geospatial, oil and gas, and public safety, helping to maximize productivity, enhance safety, and reduce risks.

Finden is an AI-powered workspace designed to connect, organize, and manage files and data across various digital locations, including drives, emails, messages, and applications. It enables users to search, chat, and research across all their personal and enterprise data, providing instant answers and insights. The platform aims to cut through digital noise, highlight important information, and help users make faster decisions. Key features include finding files quickly, getting instant answers through chat, organizing data, and instant recall. Finden integrates with existing apps and tools, consolidating digital workspaces into one secure, intelligent hub. It emphasizes data security with advanced encryption, rigorous compliance, and proactive threat protection, ensuring queries are sent to a private AI model and data is never stored or used for training.

happycapy provides an agent-native computer environment within your browser, enabling AI agents to perform tasks around the clock. Users can delegate work using Claude Code, leveraging access to over 150 AI models within a secure sandbox. The platform offers various plans, from a free tier with limited credits and basic sandbox access to Pro and Max tiers that include increased credits, enhanced sandbox environments (up to 4 cores, 8GB RAM, 200GB storage), automation capabilities for recurring tasks, and email access. It's designed for individuals and teams looking to automate browser-based workflows and integrate AI into their daily productivity.

Torq AI positions itself as an ultimate productivity assistant, primarily delivered through an AI Chrome extension. Its core functionalities include AI-powered browsing, which likely enhances the user's web experience by providing intelligent assistance or information. Additionally, it features a Smart Reply for Gmail, aiming to streamline email communication by generating intelligent response suggestions. The tool focuses on improving email productivity and overall browsing efficiency, making it suitable for individuals looking to leverage AI for daily digital tasks. While the exact mechanisms of its AI-powered browsing are not detailed, the emphasis is on assisting users in their online activities and communication.

Semantic Kernel is an open-source SDK designed for building intelligent AI agents and multi-agent systems. It provides a flexible framework for integrating large language models (LLMs) into applications, enabling developers to create sophisticated AI-powered solutions. The tool is model-agnostic, meaning it can work with various LLMs, and is enterprise-ready, supporting automated workflows and collaborative development. It empowers developers to build AI applications that can understand, reason, and interact with users and other systems, making it suitable for a wide range of AI development projects.

skrl is an open-source, modular Reinforcement Learning (RL) library implemented in Python, supporting PyTorch, JAX, and NVIDIA Warp. It is designed with a focus on modularity, readability, simplicity, and transparency of algorithm implementation, making it suitable for both research and development. The library supports a wide range of environment interfaces, including OpenAI Gym, Farama Gymnasium, PettingZoo, and ManiSkill. Additionally, it allows for loading and configuring NVIDIA Isaac Lab and MuJoCo Playground environments, enabling simultaneous training of agents by scopes within the same run. skrl is under active continuous development, with the latest updates available on its develop branch.

tools is an open-source project designed to empower AI agents with a comprehensive suite of capabilities. It adopts a model-driven approach, enabling developers to build sophisticated AI agents with just a few lines of code. The platform offers a wide array of ready-to-use tools, including file operations for reading, writing, and editing, secure shell integration for executing commands, and HTTP client for API requests. It also features advanced functionalities like memory management with Mem0 and Amazon Bedrock Knowledge Bases, web infrastructure tools for searching and crawling, and Python execution with state persistence. Furthermore, tools supports image and video processing, audio output, environment management, journaling, task scheduling, and advanced reasoning. Its unique offerings include swarm intelligence for parallel problem-solving, agent-as-tool for nested agent instances, and a multi-agent graph for deterministic pipelines, making it a versatile solution for complex AI agent development.

AlphaAvatar is a real-time interactive Omni-Avatar personal assistant framework designed to evolve into an intelligent personal butler. It is fully self-hostable and privacy-first, allowing deployment locally or on your own infrastructure with full control over data, memory, and behavior. Built around a plugin-based Agent architecture, AlphaAvatar combines full-modality memory, dynamic persona understanding, self-improving reflection, long-term planning & execution, external tool integrations, and real-time virtual characters. This enables it to move beyond a traditional chatbot into a continuous, personalized, and proactive assistant system, supporting text, voice, and visual interaction.

Rezonant is an AI-powered platform designed to bridge the gap between product vision and engineering execution. It understands product intent and codebase, helping teams turn ideas into actionable tasks and shipped code. Key features include the ability to understand product systems, shape intent into buildable work, and deliver code without losing control. Rezonant connects to existing product stacks like Jira, Linear, and code repositories to provide answers and create artifacts. It generates structured tasks with complete context for coding agents, which can be pushed directly to Jira or Linear. The platform also features a coding agent that explores the codebase and begins implementation in a secure, isolated environment, ensuring a seamless transition from concept to deployment.

Appified.ai is a no-code platform designed to transform OpenAI Assistants into fully functional web applications. This enables users to easily embed their AI assistants directly onto their websites, share them with others, or even commercialize them as products. The platform supports advanced features such as function calling and API integration, allowing for dynamic and interactive AI applications. A key differentiator is its focus on security, ensuring that OpenAI API keys remain private and secure. Appified.ai simplifies the deployment of AI agents, making sophisticated AI accessible to a broader audience without requiring extensive coding knowledge.

Dexter is an autonomous financial research agent designed to conduct deep financial analysis. It intelligently decomposes complex financial questions into structured research steps, autonomously executes tasks using various tools to gather financial data, and self-validates its work, iterating until tasks are complete. Dexter provides access to real-time financial data, including income statements, balance sheets, and cash flow statements. It also incorporates safety features like loop detection and step limits to prevent runaway execution, ensuring controlled and efficient research. The tool can be run in interactive mode and includes an evaluation suite for testing against financial questions, logging all tool calls for debugging and history tracking.

AgenticAI offers intelligent AI digital workers designed to transform accounting operations, particularly for New Zealand and Australian accountants and bookkeepers. Their flagship digital worker, Flynn, specializes in automating monthly, period, and annual GST and quarterly BAS compliance. Flynn prepares reviewable, reconciled GST and BAS workpapers on a reliable schedule, learning each client’s chart of accounts and applying consistent rules to reduce manual handling and rework. This frees up senior staff to focus on higher-value client work. AgenticAI's digital workers are intelligent agents that learn business context, understand industry nuances, and adapt to unique operational needs, ensuring compliance standards are supported. They are presented as an alternative to costly and high-risk outsourcing, offering 100% accuracy and secure data handling within New Zealand's borders.

DiffMem is a lightweight, Git-based memory backend specifically designed for AI agents and conversational systems. It leverages Markdown files for human-readable storage and Git for tracking the temporal evolution of memories through differentials. Unlike traditional memory systems that rely on vector databases or embeddings, DiffMem uses a Git-native retrieval agent that explores the repository via shell commands to build targeted context. This approach allows agents to query and search against a compact, up-to-date knowledge base while enabling deep dives into historical changes when needed. DiffMem is designed for long-horizon AI systems where memories accumulate over time, offering scalability without sprawl, auditability, and efficient temporal reasoning.

Floating Robotics revolutionizes greenhouse farming with smart, AI-driven robots designed to automate tedious and repetitive tasks such as harvesting, de-leafing, and lowering. These compact robotic solutions integrate seamlessly into existing greenhouse infrastructures, minimizing disruption while maximizing yields. Equipped with 3D vision, onboard Edge Computing, and advanced AI, the robots detect, analyze, and interact with crops in real time. They operate 24/7, performing tasks like harvesting by day with human oversight for logistics, and fully autonomous de-leafing by night. This automation helps growers cut costs, optimize resources, combat labor shortages, and ensure precise crop handling to reduce infection risks, contributing to food security.

Gambit AI offers a software-first, platform-agnostic intelligence stack designed to unlock the full potential of autonomous systems. It moves beyond simple automation to provide adaptive intelligence that allows machines to learn, collaborate, and adapt as a single unified system. Key features include a Behavior Builder for creating and training autonomous behaviors, adaptive intelligence at the edge for real-time operation, and an intuitive interface for monitoring and controlling missions. Gambit AI emphasizes coordination and orchestration over mere control, enabling multi-robot systems to operate across various platforms (air, land, sea) and translate human intent into collective action. The technology has been validated in defense industries and is designed for scalability across diverse applications.

Observe.AI provides enterprise-ready AI agents designed to handle real-world customer interactions across voice and chat. These agents accurately capture intent, enforce critical steps, and integrate with over 250 systems. The platform ensures continuous improvement through built-in QA and governance, offering features like VoiceAI Agents for end-to-end call automation and ChatAI Agents for seamless support. Observe.AI also includes AI Copilots to guide human agents in real-time and Conversation Intelligence to analyze every interaction for performance elevation, compliance, and quality management. It emphasizes trust, control, and transparency with guardrails, auditing, and robust data governance.

Debales AI is an autonomous logistics automation platform that leverages AI agents to streamline supply chain operations. It automates manual tasks such as reading and replying to emails, updating systems like TMS, WMS, ERP, and CRM, and handling exceptions. The platform offers specialized AI agents for email, support, SMS, and phone, designed to understand customer intent, auto-classify communications, and draft responses in your brand voice. Debales AI helps freight brokers, 3PLs, and carriers automate sales, prospecting, quoting, load building, customer service, and orchestration, leading to faster order processing, reduced manual errors, and significant time savings.

adaptive-classifier is a PyTorch-based machine learning library designed for dynamic text classification. It offers continuous learning capabilities, allowing models to adapt to new examples without catastrophic forgetting, and supports dynamic class addition at runtime without full retraining. A key differentiator is its strategic defense mechanism, employing game-theoretic approaches to protect against adversarial manipulation and ensure robust predictions. The system integrates with HuggingFace transformers and includes built-in ONNX Runtime support for 2-4x faster CPU inference, enabling zero-downtime model updates in production environments. It also features a multi-label classifier with automatic threshold adaptation and prototype memory for efficient similarity search.

Adlik is an open-source, end-to-end optimizing framework designed to accelerate deep learning inference processes. It supports both cloud and embedded environments, allowing for flexible and high-performance deployment of various deep learning models. The framework includes a Model Optimizer with components like pruners and quantizers, and a Model Compiler that supports optimizing technologies for models developed with TensorFlow, Keras, PyTorch, and more. Adlik also features a Serving Engine that provides an optimized runtime based on the deployment environment. Users can compile models from formats like H5, CheckPoint, ONNX, and SavedModel to Openvino, TensorFlow, TensorFlow Lite, and TensorRT, and then use provided Docker images for model inference.