Collected AI Tools
Estimated Reading Time: 18 minutes Last updated on 15th January 2025
AI tools refer to software applications and platforms that leverage artificial intelligence technologies to perform tasks traditionally requiring human intelligence. These tools encompass a wide range of functionalities, from natural language processing and computer vision to decision-making and predictive analytics. They are increasingly being integrated into various industries such as healthcare, finance, marketing, and manufacturing to enhance efficiency, accuracy, and productivity. AI tools can automate repetitive tasks, provide insights from large datasets, and facilitate personalised user experiences. As they continue to evolve, these tools hold the potential to transform how businesses operate and innovate, making them indispensable assets in the modern technological landscape. However, their adoption also raises important considerations around ethics, privacy, and the future of work.
Here then are some tools that I have encountered so far. While many are online, some offline tools are here too. The upheaval that AI is bringing upon us is akin to the personal computing one from around thirty years ago. Even the mobile revolution within the last ten or fifteen years does not compare so strongly in extent or depth. This episode could change everything for humanity. That is why my own dalliances with AI tooling proceed with a certain amount of impetus.
Aider is an AI pair programming tool that operates within terminal environments, enabling collaboration with Large Language Models (LLMs) to edit code in local Git repositories. It works optimally with GPT-4 and Claude 3.5 Sonnet but can connect to various LLMs. Key features include editing multiple files simultaneously for complex requests, automatic committing of changes with sensible commit messages, support for popular programming languages like Python, JavaScript, TypeScript, PHP, HTML, and CSS, using a map of your entire Git repo to work effectively in larger codebases, adding images and URLs to the chat for additional context, and enabling voice-based coding. Aider has achieved one of the top scores on SWE Bench, a challenging software engineering benchmark. To use Aider, you can install it via pip and run it within your Git repository using the provided command line instructions. Many users have reported improved coding efficiency and tackling complex tasks after adopting this tool.
ChatGPT is an advanced AI language model developed by OpenAI that uses deep learning techniques, particularly transformer architecture, to understand and generate human-like text based on user inputs. It excels at natural language understanding and contextual awareness, enabling intuitive and human-like conversations. With its ability to create various types of content, including articles, essays, poetry, and code snippets, ChatGPT is a versatile tool for applications in customer service, content creation, education, software development, and healthcare. The latest iteration introduces multimodal features that allow processing across different formats like text, audio, and images. It also has capabilities such as automatic summarisation, sentiment analysis, and a wide range of applications from generating engaging copy to handling inquiries in customer service or creating personalised learning experiences. As AI technology continues to evolve, ChatGPT’s potential uses are expected to expand even further, enhancing productivity across numerous fields.
Claude is an AI assistant developed by Anthropic, a company led by Daniela and Dario Amodei. Launched in March 2023, it specialises in text-based conversations, excelling at tasks like summarisation, editing, Q&A, decision-making, and code-writing using large language models with a context window of up to 200,000 tokens (150,000 words). Powered by over 137 billion text and code parameters, it can analyse text, images, charts, graphs, and technical diagrams. Anthropic’s “constitutional AI” approach aligns Claude with principles like freedom and privacy. Despite limitations such as the inability to access the internet for current events or generate images, Anthropic continues developing Claude to compete with other leading AI assistants, having raised significant funding including a partnership with Google.
CodeConvert is an AI-driven tool designed to facilitate code conversion across more than 25 programming languages, including commonly used ones such as Python, Java, C++ and JavaScript. The platform features a straightforward interface that enables developers to convert code with a single click, without needing to create an account or provide payment details for basic functionality. Key characteristics of the tool include its AI-powered conversion capabilities, ease of use, and extensive language support. It offers a free version with no registration required, maintaining user privacy by not storing inputs or outputs. For those seeking more advanced features, CodeConvert provides paid options that allow unlimited usage, custom conversion rules, and integration with popular development environments and code repositories. The tool is continually updated to incorporate the latest programming languages and enhancements, serving as a useful asset for developers aiming to optimise their coding workflow.
Codeium is an AI-powered coding assistant aimed at boosting developers’ productivity and efficiency. It offers intelligent code suggestions in over 70 programming languages and features an in-editor AI chat for seamless assistance. The tool simplifies the search for code snippets and integrates with over 40 editors, such as Visual Studio Code, JetBrains and Jupyter. Prioritising data privacy, it employs end-to-end encryption without training on user data. Codeium is available in various plans, including a free option for individuals, a teams plan with added features and an enterprise option with on-premises deployment.
ComfyUI is an open-source, node-based application for generating images from text prompts. Developed by comfyanonymous and released in January 2023, it has gained significant traction within the AI art community due to its integration with diffusion models like Stable Diffusion. Key features include a visual workflow system, support for multiple diffusion models, saveable workflows in JSON format, and customisable extensions. The software has over 50,000 stars on GitHub and is supported by Nvidia’s RTX Remix modding software, as well as the Open Model Initiative. Despite some learning curve and performance challenges, ComfyUI offers flexibility and visual clarity that make it a popular tool for artists and developers interested in AI-generated imagery.
FauxPilot is an open-source, self-hosted tool designed as an alternative to GitHub Copilot, focusing on AI-assisted code generation while maintaining privacy by running locally. It leverages SalesForce CodeGen models within NVIDIA’s Triton Inference Server to offer smart code suggestions. Unlike GitHub Copilot, it does not use OpenAI Codex, potentially reducing licensing issues and copyright risks. Users can customise and train their own AI models with specific code bases, allowing greater control over the output. Additionally, it serves as a research platform for developing and assessing code models aimed at producing more secure code. FauxPilot is suited for developers and organisations prioritising control over their coding tools and safeguarding data privacy and security, requiring specific hardware such as an NVIDIA GPU with Compute Capability of at least 6.0 and adequate VRAM.
Figstack is an AI-driven tool aimed at boosting developers’ productivity by simplifying the coding process across various programming languages. It provides features such as translating complex code into simple English, converting code between over 70 programming languages, generating comprehensive function documentation, analysing code efficiency with Big O notation, and integrating with GitHub for version control and collaboration. Figstack is available in multiple pricing tiers, with a free plan offering limited credits and paid plans providing more extensive features. Its user-friendly interface makes it suitable for developers of all experience levels, enabling them to concentrate on complex programming tasks by automating routine development aspects, ultimately enhancing efficiency and productivity.
GitHub Copilot is an AI code completion tool developed by GitHub and OpenAI, launched in October 2021. Supported by popular IDEs like Visual Studio Code, JetBrains IDEs, Neovim, and primarily Python, JavaScript, TypeScript, Ruby, and Go, it assists developers with generating real-time suggestions, error correction, documentation assistance, unit testing, code optimization, security features, increased productivity, accelerated learning, better collaboration, and is used by over 77,000 businesses worldwide. Future developments include the integration of a chatbot based on GPT-4 in March 2023 to enhance user interaction through voice commands and more sophisticated conversational capabilities. GitHub Copilot represents significant advancements in software development tools, improving efficiency, reducing repetitive tasks, and enhancing overall code quality.
Google’s DeepMind has developed a multimodal AI suite, Google Gemini, capable of processing various formats such as text, images, audio, and video. With an expanded context window of up to 1 million tokens, Gemini excels in understanding complex queries by handling multiple inputs simultaneously. It is designed for seamless integration with Google applications like Gmail and Docs and supports 35 languages. Capabilities include advanced reasoning abilities, image/video processing, custom chatbots (Gems), and features such as “Audio Overviews” that generate spoken summaries from text input. Comparatively, Gemini surpasses models like OpenAI’s GPT-4o due to its larger context window and native multimodal capabilities, making it a leading competitor in generative AI applications.
Hugging Face is a leading machine learning platform founded in 2016 by Clément Delangue, Julien Chaumond, and Thomas Wolf. Known for its Transformers library, the company simplifies access to pre-trained models for natural language processing tasks like text classification, summarisation, translation, and question answering. With over 900,000 models and 200,000 datasets available on the Hugging Face Hub, users can share, and experiment with, various machine learning applications. The platform fosters a strong community and offers extensive documentation, tutorials, and an Inference API for both free prototyping and paid production workloads. By democratising access to machine learning tools, Hugging Face continues driving advancements in artificial intelligence technologies.
Ideogram is a user-friendly, text-to-image AI generator that enables users to create visually appealing images from written prompts. The tool caters to various users, including artists, marketers, and content creators, who can generate photorealistic images and artwork quickly and efficiently using deep learning neural networks. Users can input text prompts which the AI translates into corresponding visuals, offering customisation options such as styles and moods. All in all, Ideogram is a versatile tool that facilitates creativity and streamlines the design process across various fields.
Inquisite is a research platform powered by AI technology from Duke University professors, designed for efficient information gathering and document generation. Its proprietary engine searches, reads, ranks sources, and synthesises data into coherent reports or presentations while providing citations. Collaborative tools facilitate group projects. Applicable across academia, business analysis, and content creation sectors, Inquisite offers a free tier with additional features in paid plans. By reducing time spent on manual searches, it aims to improve research efficiency while ensuring credible information.
Meta’s Llama (Large Language Model Meta AI) is a series of advanced open-source language models developed by Meta, designed to facilitate various natural language processing tasks and promote accessibility in the AI community. The Llama models include variants like 3.1 and 3.2, featuring configurations ranging from 8 billion to 405 billion parameters and support for extended context lengths of up to 128K tokens. They can process both text and visual inputs, enabling applications in chatbots, content creation, data analysis, education, and more. Trained on over 15 trillion tokens from diverse sources, these models provide developers with powerful tools to innovate across various applications while ensuring safety through features like Llama Guard and Prompt Guard. Accessibility is emphasised through open-source availability under certain licensing constraints and partnerships with cloud providers for broader accessibility. The Llama models represent a significant advancement in open-source AI technology, offering flexibility, multimodal capabilities, and community engagement.
LLaVA, an open-source multimodal AI model developed through a collaboration between researchers from the University of Wisconsin-Madison, Microsoft Research, and Columbia University, uses a transformer architecture to integrate language and visual understanding capabilities. This advanced system can process queries involving both images and text simultaneously, enabling users to discuss image content. Trained on a relatively small dataset compared to other models, LLaVA demonstrates impressive performance in various benchmarks, particularly those requiring deep visual understanding and instruction-following. The model utilizes the CLIP (Contrastive Language–Image Pre-training) visual encoder for processing images, allowing it to bridge the gap between text and images effectively. LLaVA has undergone several iterations, with recent developments focusing on improving vision-language connectors and data tailored for academic tasks. This open-source alternative to established models like OpenAI’s GPT-4 is a valuable tool for various applications due to its ability to seamlessly integrate language and vision.
LightLLM is a Python-based Large Language Model (LLM) inference and serving framework designed for high performance and scalability. It features efficient GPU utilization through tri-process asynchronous collaboration, efficient handling of requests with large length disparities using Nopad attention operations, dynamic batch scheduling, and faster inference and reduced GPU memory usage through FlashAttention integration. Additionally, it offers multi-GPU token parallelism, optimised GPU memory management via Token Attention, high-performance Router for system throughput optimization, and an Int8KV Cache for increased token capacity (Llama models only). LightLLM is based on open-source implementations like FasterTransformer, TGI, vLLM, and FlashAttention. It’s lightweight and scalable, making it suitable for deploying and serving large language models efficiently. Benchmarks show competitive performance with other frameworks on various platforms. The use of OpenAI Triton for kernel implementation and its relatively compact codebase make it easier for developers to optimise the framework. Instructions to launch and evaluate LightLLM’s performance are provided in its documentation.
Microsoft 365 Copilot is an AI-integrated productivity tool within Microsoft Office Suite that uses advanced capabilities like GPT and DALL-E for task assistance across various applications, including Word, Excel, and PowerPoint. The user-friendly interface allows interaction via voice commands or chat interfaces, enhancing the experience while automating routine tasks to save time. Copilot is available on multiple platforms such as Windows and Microsoft Edge for versatility in various work environments. It aims to boost productivity by providing intelligent suggestions, allowing users to focus on more strategic aspects of their jobs. The integration within familiar applications simplifies usage and access to AI capabilities. In conclusion, Copilot is a valuable addition to the modern workplace that empowers individuals with enhanced capabilities and streamlined workflows.
Mistral is a sophisticated AI platform developed by a French startup that offers open-weight, customisable models like Mistral 7B and Mixtral 8x7B for various applications across industries. The platform boasts exceptional speed, efficiency, and top-tier reasoning capabilities in multiple languages. It supports serverless APIs, public clouds (such as Azure and Amazon Bedrock), and on-premise setups, ensuring flexibility for users. Mistral recently introduced the Pixtral Large Model with 124 billion parameters, excelling at both text and image processing, making it suitable for complex tasks like document question answering and chart analysis. The platform’s Le Chat interface has been updated with web search capabilities, a collaborative “Canvas” tool, and advanced document/image understanding features to enhance user productivity. Mistral provides flexibility through various deployment options and community support while catering to developers, businesses, and researchers with high-performance models. However, it may come with an initial setup complexity and learning curve, as well as potential cost considerations for smaller users. The platform’s combination of high performance, flexibility, and community-driven support makes it appealing to both startups and established enterprises.
NotebookLM, developed by Google, is an advanced AI-driven note-taking and research tool that uploads, summarises, and interacts with documents in various formats. It provides a platform for organizing notes, extracting key information through Q&A sessions, creating outlines, generating study guides or FAQs, and producing audio summaries. Notable benefits include time efficiency, enhanced productivity, and improved information retention. Designed for students, educators, and professionals alike, this innovative technology streamlines research processes and adapts to individual needs.
Ollama is an open-source tool designed for running large language models (LLMs) locally on personal devices, catering to developers, researchers, and businesses seeking enhanced control over their AI applications while addressing privacy concerns. Key features include local execution of various LLMs, ensuring data security by keeping sensitive information on the user’s machine, model customisation, extensive library access, a user-friendly interface with command line functionality, and seamless integration with programming languages and frameworks. Ollama is versatile, applicable in numerous scenarios such as chatbots, summarisation tools, creative writing aids, and more. It offers significant advantages through its robust solution for local AI model execution, focusing on providing a user-friendly environment for managing large language models.
Perplexity AI is an innovative search engine powered by artificial intelligence (AI) that offers conversational search capabilities, multimodal content generation, user-centric design, and real-time information through a natural language interface. Founded by Aravind Srinivas and his team, the platform aims to prioritize users over advertisers, providing a comprehensive search experience. Perplexity AI uses large language models (LLMs) to interpret queries contextually and generates answers in a conversational format with citations and suggestions for further refinement. It can also create various types of content like text, images, and simple code by integrating multiple LLMs and image generation models. The platform is designed to be intuitive and user-friendly, with features such as swiping to delete threads and voice-based interaction on mobile devices. Perplexity AI indexes the web daily, allowing it to provide up-to-date information. Users can create a free account for additional features like sharing collections and customising preferences. The platform supports search threads and offers a Focus feature for direct searches across various categories.
Privy is an AI-powered coding assistant focused on enhancing developer productivity while maintaining a strong emphasis on privacy and security. It is open-source and self-hosted, granting users control over code and data. Privy supports multiple programming languages and integrates with popular IDEs such as Visual Studio Code. It offers features like AI-driven code completion, generation, and in-editor chat assistance. Additional capabilities include code explanation, documentation generation, unit test creation, and bug detection. Compatible with various AI models like DeepSeek Coder, CodeLlama, and Mistral, users can tailor the tool according to their hardware and preferences. Designed for both individual developers and teams, Privy serves as a comprehensive alternative to proprietary AI coding assistants by promoting open-source principles and user data control.
Refraction is an AI-powered tool that aids developers in optimising their software development by offering an extensive range of features across numerous programming languages. It supports 56 languages, including Python, Java, JavaScript and C++, providing capabilities like AI-driven code generation, automatic unit test creation, code refactoring, and inline documentation. Integrated with Visual Studio Code, it allows users to generate code snippets, create unit tests, refactor existing code, automatically document code, add debugging statements, and convert code between different languages. The tool aims to reduce the time spent on repetitive tasks, empowering developers to dedicate more attention to intricate programming challenges. It is available in both free and paid versions, catering to individual developers and teams striving for greater coding efficiency.
Tabby is an open-source, self-hosted AI coding assistant designed to enhance developer productivity with real-time code suggestions and autocompletion for various programming languages. It provides a self-hosted solution that allows developers to maintain control over their code and data, offering an alternative to proprietary tools. Tabby seamlessly integrates with popular development environments such as Visual Studio Code, Vim/Neovim, and JetBrains IDEs and is compatible with major coding large language models, enabling the combination of preferred models. It emphasises end-to-end optimisation for efficient code completion while ensuring privacy and security by eliminating the need for external database management systems or cloud services. Driven by a community-focused approach, it engages users through various platforms, making it a flexible, secure, and efficient coding assistant for individual developers and teams.
Tabnine is an AI-powered code completion assistant intended to boost developers’ productivity. Its key features include real-time intelligent code suggestions, support for over 80 programming languages and frameworks including JavaScript, Python, Java, and C++, and integration with major IDEs like VS Code, Eclipse, and JetBrains products. It offers context-aware completions relevant to the current project and coding context, can translate natural language queries into code snippets, and prioritises security with SOC-2 compliance and options for local adaptability. Tabnine provides three pricing tiers: a free Starter tier with basic completions, a Pro tier with advanced features and an Enterprise tier with custom pricing tailored for large organisations focusing on enhanced security and customisation. In summary, Tabnine seeks to expedite software development, ensure code consistency, and aid in code reviews while upholding privacy and security.
TLDR This is an AI-powered summarisation tool designed to help users quickly digest lengthy articles, documents, and other text-based content by condensing key points into concise summaries using advanced natural language processing algorithms. Users can customise the length of their summaries, making it adaptable for various needs, and access it as a web application or browser extension to simplify information gathering from favourite websites. The tool’s applications include academic research, professional efficiency enhancement, and staying informed about current topics. In short, TLDR This offers automatic summarisation capabilities and API access for developers.
You.com is an AI-driven search engine and productivity platform launched in 2021 that focuses on user customisation, privacy, and advanced capabilities. It offers a free tier for basic features and a Pro subscription with unlimited queries and extra features. The platform caters to knowledge workers, content creators, and general users by providing personalised AI-powered search results and interactive experiences. Despite competition, its emphasis on customisation sets it apart in the market, potentially leading to future developments that enhance productivity across various sectors.