AI Agents & Automation

KLINX is an AI-powered Fintech solution designed to improve back-office efficiency by simplifying incoming payment reconciliation. It leverages smart digital matching and accounting to automate the process, significantly reducing cost and time associated with reconciling receipts. The platform provides comprehensive governance over the entire reconciliation workflow, ensuring timely accounting of incoming payments, monitoring of financial flows, and effective credit risk management. KLINX also assists in the timely identification of unpaid accounts receivable and facilitates the activation of credit recovery processes for overdue invoices. It has been successfully adopted by major financial institutions like Fiditalia, Ifitalia, Fire, BNL, OAM, UniCredit, and Air Liquide.

prismatic-vlms offers a flexible and efficient codebase for training visually-conditioned language models (VLMs). It natively supports diverse visual backbones like CLIP, SigLIP, and DINOv2, with an easy mechanism for adding new ones via TIMM. The tool also integrates with arbitrary instances of AutoModelForCausalLM from Transformers, including both base and instruct-tuned language models. Designed for easy scaling, prismatic-vlms leverages PyTorch FSDP and Flash-Attention to efficiently train models ranging from 1B to 34B parameters on configurable dataset mixtures. It also includes an evaluation codebase for rigorously testing VLMs across 12 vision-and-language benchmarks and provides full instructions and configurations for reproducing results.

pytorch-attention offers a robust PyTorch implementation of various cutting-edge deep learning models, including a wide array of attention mechanisms, vision transformers, MLP-like models, and convolutional neural networks. This open-source codebase is designed for researchers and engineers to easily experiment with and integrate advanced architectures into their projects. It features implementations of models like Squeeze-and-Excitation Attention, ViT, ResNet, and MLP-Mixer, complete with code examples for quick setup and testing. The repository is modular and extensible, making it a valuable resource for anyone working on computer vision and deep learning tasks, providing a foundation for both academic research and practical application development.

LikeClaw AI Agent & Automation is a cloud-native platform designed to provide secure and efficient AI agent capabilities, addressing the security and setup challenges of tools like OpenClaw. It allows users to run AI agents in isolated, secure sandboxes, ensuring that tasks execute without compromising local machines. The platform features over 100 AI models from various providers, a vetted skills marketplace to prevent malicious scripts, and a credit-based pricing model with no subscriptions. Users can sign up and run their first AI agent task in under 60 seconds, making it accessible for individuals, freelancers, and small teams seeking autonomous AI with built-in guardrails and cost controls.

repo2txt is a web-based tool designed to convert the contents of GitHub repositories into a single, formatted text file. This is particularly useful for AI-assisted development and preparing prompts for Large Language Models (LLMs). The tool offers multiple sources including public and private GitHub repositories with token support, local file directory selection, and zip file uploads. It features smart filtering options like extension filters, .gitignore support, custom patterns, and directory selection, all previewed with a visual file tree. Performance is optimized with virtual scrolling, code splitting, web workers, progressive loading, and smart caching. It also boasts a modern UX with dark mode, responsive design, real-time GPT token counting, and privacy-first processing that is 100% browser-based with no server uploads or tracking.

OptiMate is an open-source collection of libraries developed by Nebuly AI, aimed at optimizing AI model performance. While it is now in a legacy phase and no longer actively maintained, the source code remains available for reference. Key components include Speedster, which helps reduce inference costs by leveraging state-of-the-art optimization techniques for AI models on various hardware, and Nos, designed to lower infrastructure costs through real-time dynamic partitioning and elastic quotas for Kubernetes GPU clusters. Additionally, ChatLLaMA is included for fine-tuning optimization and RLHF alignment to reduce hardware and data costs. The project is ideal for developers and data scientists looking to explore or implement AI model optimization techniques.

Kiso Technology specializes in Applied AI, engineering solutions for significant impact across various industries. They develop algorithmic software to power autonomous planning and logistics in sectors such as data centers, semiconductors, grocery, and retail. Beyond building software, Kiso Technology also works directly within organizations to construct AI applications that enhance the bottom line and help clients capitalize on transformation opportunities. Their team comprises experts in AI, operations research, machine learning, data science, and software engineering, bringing deep technical expertise to solve complex business challenges and deliver measurable impact through AI.

PINNs-Torch is a PyTorch-based implementation of Physics-Informed Neural Networks (PINNs), designed to accelerate scientific computing tasks. A key differentiator is its integration of CUDA Graphs and JIT Compilers (TorchScript), which can boost performance by up to nine times compared to earlier TensorFlow v1 implementations. The package is open-source and provides a robust framework for researchers and developers to build and experiment with PINNs. It includes examples for various problems, such as the Navier-Stokes PDE, and offers flexible installation options for both users and contributors. The tool is ideal for those looking to leverage the power of PyTorch for physics-informed machine learning, with a focus on speed and usability.

LongNet is an open-source implementation of the plug-in and play attention mechanism described in the paper "LongNet: Scaling Transformers to 1,000,000,000 Tokens." This Transformer variant is designed to significantly extend the sequence length that models can handle, reaching up to 1 billion tokens, while maintaining strong performance on shorter sequences. Its core innovation is dilated attention, which expands the attentive field exponentially as the distance between tokens grows. LongNet offers linear computational complexity and a logarithmic dependency between tokens, making it suitable for distributed training of extremely long sequences. Its dilated attention can be seamlessly integrated into existing Transformer-based optimization methods, providing a drop-in replacement for standard attention.

Intelous is an AI-driven platform designed to revolutionize B2B marketing and sales through full-funnel Account-Based Marketing (ABM) solutions. It leverages conversational AI to engage leads in personalized 1:1 dialogues, nurturing and qualifying them throughout the sales pipeline. The platform offers key features like precise buyer intent data, tailored content delivery for specific accounts, and demand cultivation to engage high-quality contacts. Intelous aims to increase meetings booked, boost conversations with target accounts, and enhance contract value. It provides solutions for both marketing teams to transform MQLs into high-quality leads and sales teams to engage, validate, and qualify prospects, turning outbound efforts into inbound opportunities. The platform also includes a Data Studio with over 200 million verified contacts and Full Funnel Avatars for sales automation.

PromptOS is an intelligent platform designed to help users create optimized system prompts. It focuses on streamlining the prompt creation process, making it easier for prompt engineers and AI users to develop effective prompts. The platform aims to provide a smart environment for crafting prompts that yield better results from AI models. While specific features are not detailed on the available pages, the core offering revolves around enhancing the quality and efficiency of prompt engineering. This tool is ideal for individuals and teams looking to refine their interactions with AI systems through well-structured and optimized prompts.

Mini-Agent is a minimal yet professional open-source demo project designed to illustrate best practices for building AI agents using the MiniMax M2.5 model. It leverages an Anthropic-compatible API, enabling interleaved thinking to enhance M2's powerful reasoning capabilities for long and complex tasks. Key features include a full agent execution loop with basic file system and shell operations, persistent memory via an active Session Note Tool, and intelligent context management that automatically summarizes conversation history for infinitely long tasks. The project also integrates 15 professional Claude Skills for documents, design, testing, and development, and natively supports MCP for tools like knowledge graph access and web search. With comprehensive logging and a clean CLI, Mini-Agent serves as an excellent starting point for advanced agent development.

Awesome Machine Learning On Source Code (MLonCode) is a curated list of research papers, datasets, and software projects focused on the application of machine learning to source code. This repository provides a valuable resource for researchers and practitioners interested in topics such as program synthesis, source code analysis, neural network architectures, and embeddings in software engineering. While the repository is no longer actively maintained, it offers an extensive collection of historical and foundational works in the MLonCode domain, including digests, conference proceedings, competitions, and talks. It covers various sub-fields like program translation, code suggestion, bug detection, and code optimization.

SnakeAI is an open-source project designed to train a neural network to play the classic game Snake through the application of a genetic algorithm. Each snake within the simulation is equipped with a neural network, initially featuring an input layer of 24 neurons, two hidden layers of 16 neurons, and an output layer of 4 neurons. A key feature is the ability to customize the number of hidden layers and neurons, offering flexibility for experimentation. The snake's 'vision' system provides 24 inputs by detecting the distance to food, its own body, and walls in 8 directions. The evolutionary process involves natural selection across generations of 2000 snakes, with fitness scores determining reproduction. Snakes are rewarded more for higher scores than simply staying alive, with a move limit to prevent endless looping. Crossover and mutation mechanisms are used to evolve the neural networks, and models can be saved and loaded for further testing and analysis.

Juicy AI offers an open AI character platform where users can engage in unfiltered NSFW chat with AI companions. The platform allows for free use, enabling users to interact with AI girlfriends or boyfriends, including sexting. A key feature is the ability to create custom anime characters, providing a personalized experience. It is designed for entertainment purposes, focusing on unrestricted AI interactions and character creation within an adult-oriented context. The tool emphasizes a no-filter approach, making it suitable for users seeking explicit AI conversations and character customization.

JazzX AI is an AI execution platform built specifically for the mortgage industry, aiming to automate and optimize the entire mortgage lifecycle. It goes beyond generic AI tools by providing governed, end-to-end automation that leads to audit-ready decisions, faster cycle times, and reduced costs per loan. The platform ensures every decision is backed by full-file reasoning with traceable, source-cited findings, eliminating idle queues and compressing cycle times. JazzX AI also focuses on accelerating pipeline conversion through a guided borrower and loan officer experience, and unifies various personas, workflows, and data layers above existing systems to prevent tool sprawl.

Iternal Technologies offers enterprise AI solutions focused on secure, local, and accurate AI deployment. Key products include AirgapAI, which provides 100% local, air-gapped AI for sensitive data and regulated industries, and Blockify, a patented technology that achieves 78x greater accuracy by preventing AI hallucinations through multi-layered validation. The platform also features AI Operations & Orchestration with Nebulous, partner enablement with PRISM, and extensive AI education through AI Academy. Iternal aims to eliminate multi-vendor complexity and accelerate AI adoption, serving industries like defense, healthcare, and financial services.

SmolLM 360M Instruct WebGPU is a powerful and fast AI chatbot that operates locally, allowing users to generate text by providing prompts. This application enables you to input a starting sentence or topic, and it will create a detailed response or continuation. The tool is hosted on Hugging Face Spaces, which offers various hardware options for running applications, including free CPU instances and paid GPU options. It is ideal for individuals and teams looking for a responsive AI text generation solution that can be deployed and managed within the Hugging Face ecosystem.

Powered By Intel Leaderboard is a platform designed for the evaluation and comparison of open-source language models (LLMs) specifically on Intel hardware. Users can submit their LLMs to be ranked on the leaderboard, providing details such as the model name, the Intel hardware used for evaluation, and the precision settings. This tool offers a valuable resource for developers and researchers to gain visibility for their models and receive feedback on performance. It serves as a centralized hub for understanding how different LLMs perform within the Intel ecosystem, fostering innovation and development in the AI community.

NeuralNLP-NeuralClassifier is an open-source toolkit designed for implementing neural models for hierarchical multi-label text classification. This tool addresses the complexities of real-world classification tasks by offering a variety of text encoders, including FastText, TextCNN, TextRNN, RCNN, VDCNN, DPCNN, DRNN, AttentiveConvNet, and Transformer encoder. Beyond hierarchical multi-label classification, it also supports binary-class and multi-class text classification. Built on PyTorch, NeuralNLP-NeuralClassifier demonstrates performance comparable to reported results in academic literature, making it a robust solution for researchers and engineers in natural language processing.

RAG-Survey is a comprehensive GitHub repository dedicated to collecting and categorizing research papers related to Retrieval-Augmented Generation (RAG) within the context of AI-Generated Content (AIGC). It offers a structured taxonomy that breaks down RAG into its foundational concepts, various enhancement techniques, and diverse applications. The repository serves as a valuable resource for researchers and practitioners, providing an organized overview of the rapidly evolving RAG field. It is actively maintained and updated to incorporate new papers and advancements, ensuring its relevance and utility for staying current with the latest developments in RAG for AIGC.

Ovis2 16B is an AI chatbot developed by AIDC-AI, available as a Hugging Face Space. This application enables users to interact with an AI by submitting text prompts, images, or videos to generate responses. The tool is designed with capabilities to "see, read, and reason," allowing it to provide detailed answers and explanations based on the input provided. While the description highlights its ability to understand and respond to various media types, the current live website indicates a runtime error, suggesting the application may not be fully operational at this time. It aims to offer a comprehensive conversational AI experience.

IoT Worlds is a comprehensive platform dedicated to the Internet of Things (IoT) and emerging technologies, including AI and Cybersecurity. It offers a rich repository of expert insights, in-depth analysis, and professional services designed to help individuals and organizations discover, learn, develop, and succeed in these rapidly evolving fields. The platform covers a wide range of topics from smart devices and industrial IoT to AI agents and cybersecurity threats, providing valuable information for both technical and non-technical audiences. With a focus on facilitating life with AI IoT technologies, IoT Worlds aims to make the world a better and more peaceful place through technological advancement and collaboration.

DeText is an open-source deep neural text understanding framework developed by LinkedIn, designed for various Natural Language Processing (NLP) tasks including ranking, classification, and language generation. It utilizes deep neural networks for semantic matching to understand user intents in search and recommender systems. The framework offers high flexibility with customizable model architectures, support for multiple text encoders (CNN, BERT, LSTM), and various data input types. It features end-to-end training, interaction modeling between ranking sources and targets, and automatic feature extraction. DeText is easy-to-use with configuration-based modeling and supports integration with tf-ranking for LTR loss and metrics.