AppsFor people who build with AI

New AI apps for builders.

New AI apps for builders, newest launch first, by category. Each listing is written from the app's own site. Sponsoring never changes the order.

Newest launch first · all categories · sponsoring never changes the order

mdkit

A personal library for organizing, syncing, and versioning reusable AI skill.md files from various AI tools.

This application helps users keep their skill.md files organized, synced, and ready for use. It enables managing versions, accessing content from the menu bar, and sharing with teams. The tool is designed for files generated by AI tools such as ChatGPT, Claude, and Notion, facilitating standardized AI output and version control for AI-related coding skills.

Sikemux

A whole team of AIcoders, in one window.

Sikemux is a native application that integrates multiple AI coding agents such as Claude Code and Codex into a single window. It allows users to manage their existing AI model subscriptions and orchestrate these agents to collaborate on coding tasks. The application enables agents to interact with a browser, perform code changes, and provides tools for reviewing modifications and integrating with developer workflows.

Even Keel

A task queue designed for managing coding agents.

Even Keel provides a task queue infrastructure for coding agents. It includes functionalities for managing agent plugins and integrating with various AI models. The system helps in orchestrating and executing tasks within an AI development workflow.

xcontrol

A small wrapper around tmux that opens and resumes your CLI agent sessions and lets agents communicate with each other.

xcontrol is a tmux wrapper designed to manage command-line interface (CLI) agent sessions. It facilitates opening, resuming, and orchestrating AI coding assistants like Codex, Copilot, and Claude within a unified tmux environment. The tool enables agents to communicate by role, handles project folders, and resumes CLI conversations without requiring additional services or Python packages.

Whiz.pub

An AI-native blogging platform enabling content publishing via API, CLI, or directly from AI agents, transforming markdown to static HTML.

Whiz.pub is an AI-native blogging platform focused on flexible content publishing. It allows users to publish markdown files by utilizing an HTTP API, a command-line interface, or by configuring their AI agents. The platform transforms incoming content into static HTML for readers.

tts-server-proxy

An Android system TTS engine that uses self-hosted Qwen3-TTS or other OpenAI-compatible speech endpoints for voice generation.

tts-server-proxy integrates self-hosted Qwen3-TTS or other OpenAI-compatible speech endpoints as a native Android Text-to-Speech service. It allows any Android application to utilize custom AI-generated voices running on a user's local GPU server. This tool provides a private, customizable speech synthesis solution without relying on cloud services or incurring per-character costs.

Search2o agent platform

A platform for developers to build, validate, and deploy task-focused AI agents that automate business workflows.

The platform enables developers to build, validate, and publish task-focused AI agents for business workflows. It provides a complete development environment including AI-assisted drafting, a schema-aware code editor, and tracing tools. Agents can interact in conversations, leveraging short-term and long-term memory, and integrate with various enterprise systems and large language models.

Sinaps Studio

A platform offering tools and an academy to learn about building, governing, and operating AI agents based on large language models.

Sinaps Studio provides tools for AI agent governance, allowing users to set access, rules, and supervision for their agents. It also enables teams to collaborate and organize operational responsibilities. The platform helps users build operations and track agent costs, results, and traceability. The Sinaps Academy offers a comprehensive course on understanding LLMs, configuring agents, and implementing governance.

eso-box

Turns a spare laptop into an always-on server for Claude Code sessions accessible from any device.

Eso-box enables users to carry their Claude Code sessions across multiple devices. It converts a spare Linux or macOS machine into an always-on server for Claude Code. Users can access their live coding sessions from laptops via SSH or from phones through the Claude app, allowing seamless transitions between devices.

Fardesk

Every Claude Code session on all your computers, on your phone.

Fardesk provides an iPhone and Mac app for managing Claude Code sessions across multiple computers. It allows users to monitor ongoing sessions, respond to questions, and initiate new coding sessions remotely. The app operates over a user's private Tailscale network, ensuring no cloud services are involved in managing the AI interactions.

Taster

Taster checks AI agent tool calls against plain-English rules before running and screens agent responses for prompt injection.

Taster checks every tool call an AI agent makes against predefined rules written in plain English before execution. It also screens what comes back from tools for prompt injection before the agent reads it. This tool aims to prevent unwanted agent actions and maintain control over agent behavior.

Synapse

A local code graph for AI IDEs, benchmarked against grep.

Synapse provides functionality for code analysis, including resolving symbol call targets and preserving call qualifiers. It supports features like typed-receiver calls and nested functions for languages such as Swift. The tool also includes a benchmark harness for performance evaluation.

Confidential Inference

Compare providers running AI inference inside trusted execution environments, filtering by models, pricing, and API features.

This product offers a directory for comparing AI inference providers. It enables users to filter and evaluate providers based on criteria such as supported models, pricing, and API features. The platform focuses on services that run AI inference within trusted execution environments, detailing their privacy, attestation, and technical specifications.

ModelRefs

ModelRefs helps builders choose, compare, and implement AI models, providers, benchmarks, workflows, and tools with evidence-based insights.

ModelRefs provides an AI reference layer for implementation decisions. It assists builders in selecting, comparing, and deploying AI models, providers, benchmarks, and workflows by offering evidence, constraints, and trade-offs. The platform offers structured model, provider, benchmark, and workflow data, along with an Evidence API for programmatic access to citations and benchmark runs.

LLM & agent cost calculator

Calculates the cost of AI agent runs and API workloads across 45 LLM models, including prompt caching and batch discounts.

This tool calculates the cost of AI agent runs and API workloads. It supports 45 models from six providers, factoring in prompt caching, batch discounts, long-context tiers, and tool calls. The calculator helps builders compare expenses across different LLMs for various AI applications.

Atlas

Turns public webpages into clean, structured, and token-efficient Markdown for AI agents.

Atlas compiles public webpages into clean, structured, and token-efficient Markdown. It transforms raw HTML into semantic artifacts that software can inspect, providing a deterministic foundation from a single URL to an entire domain. This infrastructure is designed for systems that rely on web data, facilitating scraping, discovery, crawling, and batch processing.

HissingKitty

Provides AI assistants with independent data on SaaS costs, alternatives, and build-vs-buy decisions for renewal management.

This product integrates as a connector for AI assistants like Claude, ChatGPT, and Cursor. It furnishes agents with an independent record of software costs, price changes, and suitable alternatives. The tool enables AI to answer questions about whether to keep, switch, or build a SaaS tool, along with negotiation points and competitive analysis.

Crosshead

Provides working memory and persistent identity for AI coding agents to reduce repetition between sessions.

Crosshead provides working memory and persistent identity for AI coding agents such as Claude Code and Codex. It captures recent decisions, corrections, and next steps, carrying this information into subsequent sessions. This allows agents to pick up work without needing repeated explanations and to maintain continuity across different tools.

ScreenClip

Screenshot a bug, your coding agent opens the PR.

ScreenClip is a tool designed for capturing screenshots of bugs. It integrates with a coding agent to automate the process of creating pull requests. This application helps developers streamline their bug reporting directly into the development workflow.

Barnacle

Barnacle floats over a phone screen, providing AI assistance and saving information while the user scrolls.

Barnacle is an AI assistant that operates over a phone screen, offering help and interaction as users scroll. It answers questions about on-screen content, such as identifying objects, providing clues, or generating responses. The application saves and organizes all interactions and information, making it searchable and available for future follow-ups.

Takehound

AI video editor for talking-head videos

Takehound is an AI video editor designed for talking-head videos. It automates the editing process, removing dead air, retakes, and pauses, while also incorporating user-provided clips and text. The AI, called Hound, learns from user feedback and previous edits to refine its approach for future projects.

Plinth

Ask your AI for an app, get a link to share

Plinth enables users to deploy web applications by describing their needs to an AI assistant. It integrates with AI models like Claude and ChatGPT, allowing them to create and put online custom apps, ranging from simple static pages to full applications with data storage and add-ons. The service handles the technical setup, hosting, and allows for continuous updates to the application through natural language commands.

Reverie

Local memory for coding agents that learns your corrections.

Reverie provides local memory specifically designed for coding agents. It enables these agents to learn from user corrections, improving their performance over time. This tool enhances the iterative development process for AI-powered coding assistants.

Atli

Enables AI assistants to find, buy, and install travel eSIMs directly within the chat interface.

Atli is an MCP server designed to integrate with AI assistants such as Claude, ChatGPT, and Gemini. It enables these AI assistants to find, purchase, and install travel eSIMs for users directly within the chat interface. Users can ask their AI to check phone compatibility, buy specific plans, manage existing eSIMs, and track data usage, all while maintaining payment security.

Bicycle

An audio-visual interface that enhances interaction between users and their AI agents.

Bicycle provides an audio-visual interface for interacting with AI agents, designed to increase communication bandwidth. Users can talk, sketch, and write to their agents, receiving visual and auditory feedback on agent actions and understanding. This tool helps people who write software to get tasks done across various applications by streamlining their interaction with AI agents.

Revix AI

This AI code reviewer evaluates pull requests against existing repository standards and conventions.

Revix AI is an AI code reviewer that learns the patterns a team follows and applies them to every pull request. It extracts review rules directly from the codebase, providing useful insights without manual rule definition. The tool delivers comments labeled by intent and severity, along with summaries of key changes and risk levels.

Kalike

A pure Kannada language model trained from scratch, designed to read, write, and talk exclusively in Kannada.

Kalike is an open-weight base and chat model designed exclusively for the Kannada language. It was trained from random weights using a Kannada-only tokenizer and training data, without exposure to English. The model is available in base and fine-tuned chat versions, offering capabilities for text continuation and conversation in Kannada.

OpenEye

Serves vision models like YOLO, SAM 2, and depth estimation via an MCP server and API for AI applications.

OpenEye provides an MCP server for vision models, offering an API for tasks such as object detection, image description, and depth estimation. It includes a command-line interface and supports both hosted and self-hosted inference for these models. The platform aims to make various vision AI capabilities accessible for developers.

SoftGlaze AI

An MCP server that allows AI apps to interact with and manage multi-profile browser activity on a local machine.

SoftGlaze AI is an MCP server integrated into SoftGlaze Browser, a multi-profile browser for Windows. It enables connected AI applications to read browser profiles, manage proxies, run macros, and perform actions directly within web pages. The product ensures user control by requiring explicit approval for all AI-initiated changes before execution.

Solipsist

Solipsist is a markdown editor for planning, research and writing. Your agents work inside the page, on your machine.

Solipsist is a markdown editor designed for planning, research, and writing, integrating local AI agents directly within the page. It utilizes Pi, an open-source coding agent, to perform tasks such as generating plans, conducting research, drafting reports, and creating images, all on the user's machine. The tool operates on local files, allowing agents to use connected models and tools while providing full control over the AI-generated content.