Business Software for TypeScript

OrcaRouter serves as a routing system for AI models that are compatible with OpenAI, efficiently directing prompts to the appropriate models from a wide array, including OpenAI, Anthropic, Gemini, DeepSeek, Qwen, Kimi, and over 200 other leading and open-source models. Its design aims to maintain the high quality of responses while minimizing costs associated with AI inference by evaluating each prompt and directing complex reasoning tasks to premium models while assigning simpler tasks to more economical open-source options. The routing process is meticulously quality-graded, avoiding arbitrary swaps for cheaper models, and every request clearly indicates the difficulty rating, chosen model, provider, and associated costs, ensuring that routes remain transparent, accountable, and reproducible. Developers can easily switch models by updating the API base URL, while previously established SDKs, model names, and streaming functionalities remain operational. Additionally, OrcaRouter features seamless automatic failover capabilities, allowing for traffic rerouting without interruption should a provider experience downtime, thus preventing disruptions for users. It also offers comprehensive API key management that incorporates spending limits, model allowlists, rate restrictions, and budget compliance, among other functionalities, ensuring robust control over resource usage. This combination of features makes OrcaRouter an indispensable tool for optimizing AI model utilization in various applications.

The Agent Client Protocol (ACP) serves to unify the communication between code editors, integrated development environments (IDEs), and coding agents, establishing agent-editor interoperability as a standard rather than necessitating unique integrations for every conceivable pairing. It establishes a common interface for interaction between AI agents and client applications, featuring a flexible, extensible, and platform-independent architecture suitable for both local and remote use cases. By tackling issues related to integration costs, limited compatibility, and developer dependency, ACP allows agents adhering to the protocol to function seamlessly with any compatible editor, while editors that embrace ACP can tap into a wider network of ACP-compatible agents. Much like the Language Server Protocol facilitated standardized language server integration, ACP separates agents from editors, enabling both to evolve independently, thereby empowering developers to select the most effective tools for their specific workflows. This innovation fosters a collaborative environment where tools can be easily integrated, enhancing overall productivity and efficiency for developers.

Graphify serves as an innovative open source knowledge graph engine that converts diverse inputs such as code, documentation, research papers, meetings, images, browser tabs, and commits into a single, navigable graph with full recall capabilities. Designed to function as a persistent memory for AI coding assistants, it empowers tools like Claude Code, Codex, OpenCode, Cursor, Gemini CLI, GitHub Copilot CLI, Aider, Factory Droid, Kimi Code, Kiro, Pi, and Google Antigravity with a queryable grasp of a project, thereby eliminating the need for them to continuously search through files. Users can direct Graphify to any directory, where it generates an initial corpus through AST extraction, semantic analysis, and Leiden clustering, effectively converting an entire codebase or document collection into a comprehensive graph in a single operation. Unlike traditional RAG pipelines that require re-embedding for every modification, Graphify sustains a dynamic graph that only updates the affected nodes and edges when files are altered, allowing the remainder of the corpus to remain stable even at an enterprise scale. This capability not only enhances efficiency but also facilitates seamless collaboration among various AI tools, significantly improving the overall workflow for developers and researchers alike.

Hindsight is an innovative memory framework designed to enhance AI agents by enabling them to learn progressively rather than resetting their knowledge with each new interaction. Unlike traditional memory systems that primarily focus on recalling past conversations, Hindsight prioritizes the learning process, equipping agents with a persistent long-term memory through advanced biomimetic data structures. This functionality allows AI agents to keep track of essential facts, access relevant context, and engage in reflective reasoning based on their experiences. Hindsight is particularly beneficial for agents that require a deep understanding of user identities, previous discussions, evolving preferences, decision-making histories, and necessary behavioral adjustments across different sessions. To achieve this, it incorporates three fundamental operations: retain, which captures new information; recall, which accesses appropriate memories when required; and reflect, which aids agents in synthesizing observations, developing mental frameworks, and gaining insights from earlier interactions. By implementing these features, Hindsight ensures a more personalized and context-aware experience for users.

Manufact serves as a comprehensive platform designed for the creation and deployment of MCP applications and servers, providing teams with expedited access to the ChatGPT Apps Store, Claude Connectors, and various user-agent interaction points. The mcp-use SDK functions as a complete MCP framework, facilitating the development of MCP applications for both ChatGPT and Claude, along with MCP servers tailored for AI agents. With Manufact, every phase of the MCP lifecycle is streamlined without the need for additional tools: developers can create using an SDK, a skill, or a vibe; initiate deployment with a single command; publish by following marketplace guidelines and utilizing auto-generated submission resources; refine their products through Cloud Inspector; and oversee performance with features like analytics, session replays, trace logs, error metrics, and notifications. Teams benefit from the flexibility to scaffold with the MCP-use SDK, integrate a skill into a coding agent, outline an app and observe the scaffolding process, or seamlessly incorporate an existing MCP server without modifications. Moreover, Manufact Cloud establishes a connection to a repository just once, ensuring that every push leads to automatic deployment, while providing preview URLs for pull requests, as well as managing custom domain setups and SSL certificates. This all-in-one solution enables teams to focus more on innovation rather than the complexities of infrastructure management.

Ornith-1.0 represents an innovative family of models tailored specifically for coding tasks that require agentic capabilities. This family encompasses a wide range of models, from the compact 9B Dense versions ideal for deployment on edge devices to the expansive 397B MoE frontier-scale models designed for peak performance, including variants such as 9B Dense, 31B Dense, 35B MoE, and 397B MoE. Built upon the foundational strengths of pretrained models like Gemma 4 and Qwen 3.5, Ornith-1.0 excels in achieving top-tier performance among open-source models that are similar in size when evaluated against coding benchmarks. A significant breakthrough of this model is its self-improving training framework, which effectively learns to produce both solution rollouts and the tailored scaffolds that direct those rollouts. Rather than depending on static, human-crafted harnesses, Ornith-1.0 perceives the scaffold as a dynamic entity that evolves alongside the policy, enabling the model to optimize both the orchestration of tasks and the resulting solutions in tandem. This dual optimization approach enhances the model's adaptability and effectiveness in real-world coding scenarios.

Gemini 3.5 Flash-Lite stands out as the quickest model within Google's Gemini 3.5 lineup, specifically engineered for tasks requiring low latency and for enhancing developer workflows that demand high throughput, including agentic search, document processing, coding, and extensive data analysis. It boasts an impressive output capacity of 350 tokens per second and marks a significant enhancement over earlier Flash-Lite iterations in terms of both quality and agentic capabilities. Developers have the flexibility to adjust the model's thinking level to suit the demands of the task at hand: minimal or low thinking allows for rapid processing of large volumes, while elevated thinking levels accommodate more intricate, multi-step workflows involving subagents. Furthermore, the model is equipped with built-in computational skills, enabling it to interact effectively with various digital environments across compatible platforms. Additionally, Gemini 3.5 Flash-Lite excels in coding, comprehending long contexts, and executing real-world tasks, consistently outperforming its predecessor, Gemini 3.1 Flash-Lite, in critical assessments and even exceeding the performance of Gemini 3 Flash on multiple benchmarks related to agentic functions and software engineering. This impressive performance highlights its potential to transform how developers approach complex workflows and data-intensive tasks.

Voltra is an innovative Energy Cloud Platform designed to simplify intricate physical systems via software, overseeing the comprehensive lifecycle of sites and devices from their initial setup to ongoing maintenance. At its core is Voltra Fabric, which seamlessly links devices from various manufacturers, protocols, generations, and standards using protocol-agnostic adapters, effectively creating a unified, governed graph that serves as a live digital twin and reliable reference for the site. This model enables Fabric to ensure interoperability while keeping connections, data, and configurations up to date as firmware, sites, and assets develop over time. Users can access a complete view of their site through a web browser, allowing them to track faults, usage statistics, performance metrics, telemetry, and alerts in real-time, as well as securely manage device traffic or transfer data to external data lakes and observability platforms. Additionally, developers have the flexibility to control the same infrastructure via a scriptable command-line interface and through REST and OpenAPI protocols, enabling a versatile approach to managing energy systems. This multifaceted platform empowers operators and developers alike to optimize their energy management strategies and enhance operational efficiency.

Kastra serves as the crucial authorization framework for AI systems, determining the permissions of agents, models, and tools prior to their execution. Positioned along the execution path of all interactions such as prompts, tool calls, shell commands, database operations, and API requests, it evaluates each action based on deterministic, attribute-driven policies, rendering decisions to allow, deny, redact, or escalate in less than a millisecond. In contrast to monitoring solutions that only track AI actions post-execution, Kastra proactively prevents unauthorized activities before they can impact any tool, API, database, or production environment. Its comprehensive control plane integrates a policy engine, edge decision-making capabilities, various integrations, and a tamper-proof evidence vault that securely signs each decision for auditing and replay purposes. Furthermore, with Kastra Edge, local enforcement is extended to developer environments, safeguarding coding agents such as Claude Code, Cursor, and Codex CLI from harmful commands, unauthorized data extraction, unsafe file modifications, and improper tool usage. This proactive approach to authorization not only enhances security but also ensures compliance and accountability in AI-driven processes.

Prefactor is a cutting-edge platform designed for real-time assessment, monitoring, and reliability of production AI agents. It evaluates each execution instantly based on metrics such as quality, drift, cost, and data risk, seamlessly integrating these assessments into actionable responses to ensure that any failing agent is detected in real time rather than merely reflected on a post-execution dashboard. Teams are equipped to monitor every model invocation, tool usage, and decision-making process through structured traces and spans, allowing them to conduct evaluations using LLM-as-judge, technical assessments, qualitative analyses, and custom metrics at every phase of the process. Additionally, context can be incorporated from various sources, including GitHub, Linear, Jira, databases, and internal APIs, serving as ground truth for evaluations. When a run exceeds predefined limits, Prefactor is capable of blocking or throttling it, pausing sensitive actions, or routing the decision to a person for approval, modification, or rejection prior to execution, with meticulous logging of each choice made. The command-line interface allows for the discovery of agents without the need for platform migration, while the TypeScript and Python SDKs ensure seamless integration with LangChain, Claude, Vercel AI, OpenClaw, and LiveKit, enhancing the overall functionality and adaptability of the platform. This comprehensive approach not only optimizes agent performance but also fosters collaboration among teams by providing clear visibility and control over the AI processes.

Human Browser is an innovative cloud-based solution designed specifically for AI agents, equipping them with the capability to interact with the open web via authentic Chrome sessions. Any agent that is capable of making HTTP requests can seamlessly integrate with it through A2A, while the inclusion of MCP support allows editor-based agents to connect, and a command-line interface facilitates terminal and CI workflows. These agents can be given tasks in natural language, enabling them to navigate websites, perform clicks, input text, scroll through pages, log into accounts, fill out forms, gather data, and produce structured outputs, complete with a live viewer URL that allows humans to observe the session in real time. The browser employs residential IP addresses, utilizes engine-level Chromium fingerprinting, and mimics human-like behaviors in mouse movement, typing, and scrolling. It also features advanced mechanisms to tackle challenges such as reCAPTCHA, hCaptcha, and Cloudflare Turnstile. Supporting both desktop and mobile browser profiles, it provides options for country selection from an extensive residential proxy network. Moreover, it can operate on various platforms including bare Linux, Docker, and GitHub Actions, and doesn't require a display server, VNC, or desktop environment to function effectively, making it a versatile choice for developers. This flexibility makes it an essential tool for automating web interactions in diverse scenarios.

Waterr is an innovative AI meeting platform that enables teams to assign live discussions to autonomous AI personas capable of visual and auditory interactions with participants. Users can articulate the purpose of a meeting in straightforward language or utilize pre-existing templates for various situations, including candidate screening, sales discovery, customer onboarding, coaching, feedback interviews, support, research, and ongoing operational tasks. Each AI persona can be customized with its own appearance, voice, background, and mannerisms, ensuring an experience that resembles engaging with a real human rather than a mere chatbot. Teams have the option to review the persona, scenario prompts, and the entire conversational flow, allowing them to test it live in an actual meeting before presenting it to other participants. Additionally, the same scenario can be integrated into a website as an avatar, inline widget, or a talk-to-us button, enhancing user interaction. For developers interested in leveraging this technology, Waterr offers a comprehensive API for facilitating AI meetings, complete with features such as authentication, SDKs, webhooks, rate limits, and the ability to create custom avatars for a tailored experience. This versatility makes Waterr a valuable tool for teams seeking to streamline their meeting processes while maintaining a personal touch.

Mireye serves as a comprehensive infrastructure for AI agents operating in the physical world, offering a unified API that encompasses data, enrichment, tools, and signals. Various agents, including Claude, ChatGPT, Gemini, Cursor, Grok, Copilot, DeepSeek, and bespoke systems, can utilize a single API with MCP tools to inquire about locations and receive well-cited data meticulously indexed from real-world information. Mireye is capable of resolving addresses, obtaining location-specific attributes, calculating distances, and providing the methodology behind its operations to enable result replayability. Its geocoding endpoint efficiently translates U.S. street addresses into geographic coordinates while transparently detailing the derivation of each location and rejecting low-confidence matches instead of making assumptions. The lookup function broadens a single point into contextual details such as county, census tract, congressional district, metropolitan area, timezone, elevation, and flood risk. Furthermore, the fetch capability returns specific values for requested fields or presets at various U.S. locales, with each value accompanied by its source, time of retrieval, and a confidence rating. This multifaceted API enhances the ability of AI agents to interact with the physical world in a meaningful and precise manner.

Encore serves as an automated infrastructure platform designed for both humans and AI agents, enabling developers to efficiently build, validate, and deploy applications without the delays associated with Terraform pull requests or platform tickets. By utilizing its open-source SDK, developers and agents can specify the necessary infrastructure for each feature directly within application code, while Encore autonomously provisions and verifies various resources such as databases, queues, buckets, and Pub/Sub, facilitating a seamless transition from development to production. Applications can be executed locally in an environment that closely resembles the cloud setup, enabling comprehensive testing of features and distributed systems with just a single command. When a new branch is pushed, Encore automatically generates a dedicated preview environment, allowing teams to assess workflows and external integrations independently of shared staging resources. Upon deployment, the required infrastructure is provisioned automatically within the organization’s own AWS or GCP account, VPC, and compliance parameters, streamlining the entire process. This innovative approach not only enhances efficiency but also ensures that developers can focus more on coding and less on infrastructure management.

Holo4 is H Company's series of generalist computer-use and agentic AI models built to perform multi-step work across software interfaces. It is available as Holo4 27B, a dense 27-billion-parameter model, and Holo4 35B-A3B, a Mixture-of-Experts model containing 35 billion total parameters with 3 billion active. Holo4 can interact with applications by clicking and typing through graphical interfaces, writing and executing code, or calling MCP and API tools. The same model can operate across desktops, websites, Android devices, code sandboxes, and business APIs without requiring developers to select a separate specialized model for each environment. H Company trained Holo4 using 127 billion supervised fine-tuning tokens, with approximately three-quarters consisting of successful agentic trajectories spanning desktop, web, MCP/API, and mobile tasks. Reinforcement learning then trained separate experts for desktop and web interaction and for terminal, MCP, and API work before merging them into a single model. Holo4 27B scored 85.2% on OSWorld, 61.7% on OSWorld 2.0, 45.4% on AutomationBench, and 85.1% on AndroidWorld in the evaluations reported by H Company. The models support a 256K context window, with the 27B model positioned for greater accuracy on long multi-step tasks and the 35B-A3B version positioned as a faster and less expensive alternative. Holo4 is available through a hosted API and downloadable model weights, enabling developers and enterprises to build agents that perform workflows spanning multiple applications and interaction methods.

Tev1 is a rapid decision-making and classification model developed by Together AI, which has been fine-tuned from Qwen3.5 4B. Its user-friendly interface allows for the input of a state, a question, and between 2 to 24 labeled choices, enabling the model to output a single letter that can be mapped back to a semantic key by software. This design is particularly beneficial for various production tasks, such as managing support ticket routing, assessing automated returns, enforcing policy guidelines, sentiment analysis, document categorization, and integrating precise decision-making steps into larger agent workflows. With a parameter count of 4.7B and a 32.8K-token context window, the model operates with non-thinking generation to ensure fast and reliable classification. It has been trained on a diverse set of 37,840 examples covering areas such as intent detection, boolean comprehension, sentiment evaluation, policy application, as well as categorizing news and academic papers. The recommended configuration suggests using a temperature setting of 0 along with a minimal output budget, ensuring that the responses are both consistent and straightforward to interpret. In summary, Tev1 stands out as a versatile tool that efficiently addresses a wide range of classification needs.

Span-01 is an advanced hyper-parallel reasoning classifier that identifies intricate patterns within extensive production traces in a single forward pass. Instead of producing tokens one after the other, it analyzes natural-language behavior definitions and provides direct probabilities indicating whether a behavior is present, absent, or not observable. The model undergoes initial training for general classification reasoning using Reinforcement Learning with Human Feedback (RLAIF) before it is fine-tuned specifically for behavior detection. This allows Span-01 to utilize previously unseen definitions for new behaviors and various domains, avoiding limitations imposed by a rigid taxonomy. During the inference phase, its hyper-parallel definition branches assess multiple behaviors simultaneously against the same trace context, and the incorporation of hybrid attention facilitates effective reasoning over lengthy and intricate inputs. Span-01 is meticulously crafted for monitoring production environments with a focus on safety, reliability, grounding, agent behavior, and response quality, addressing challenges such as jailbreak and prompt injection detection, as well as privacy and confidentiality concerns. Its innovative approach positions Span-01 as a significant advancement in the field, enabling more efficient and comprehensive analysis of complex behaviors.

Vitest is an innovative testing framework built on Vite, aimed at delivering rapid, lightweight testing capabilities that seamlessly align with contemporary JavaScript development practices. Leveraging Vite's architecture, Vitest is familiar with existing Vite configurations, allowing it to utilize the same resolution and transformation processes that applications employ during both development and production builds. Additionally, Vitest can operate independently of Vite and is compatible with backend code as well. It features a Jest-compatible API, which includes functionalities like expect assertions, snapshots, mocking, and coverage, facilitating an easy transition for projects moving from Jest. The framework's intelligent watch mode optimizes testing by re-executing only those tests that pertain to modified files, providing an experience akin to Vite's rapid Hot Module Replacement. Furthermore, Vitest natively supports ESM, TypeScript, and JSX, which minimizes the necessity for additional transformation setups. This combination of features makes Vitest a compelling choice for developers seeking a robust testing solution.

Boat delivers robust cloud-based virtual machines and persistent sandboxes specifically designed for managing long-lasting AI agents, empowering users to effortlessly create thousands of cloud instances equipped with web browsers and persistent storage while alleviating the burden of infrastructure management. Each sandbox functions as a comprehensive Ubuntu Linux virtual machine, providing SSH and SCP access, Docker compatibility, dedicated IPv4 or IPv6 addresses, a smooth 60 fps virtual desktop, and rapid disk-level forking capabilities from snapshots. Pre-installed tools such as Docker, VS Code, Chrome, Ghostty, GitHub CLI, Rust, Node.js, and Bun allow users to execute any development language, tool, or workflow seamlessly. Boat is compatible with various agents including Claude Code, Codex, and OpenCode, and it offers functionalities to prompt, observe, and manage agent environments remotely via its command-line interface, HTTP API, and SDKs available in both Python and TypeScript. Users can run commands, transfer files, and monitor agent communications, while also hosting applications accessible through public HTTPS links and utilizing integrated desktops and web browsers for enhanced productivity. This comprehensive solution not only simplifies the management of AI agents but also fosters a more efficient development experience for users across diverse projects.

Enhance your productivity by ensuring only high-quality code is released, as SonarQube Cloud (previously known as SonarCloud) seamlessly evaluates branches and enriches pull requests with insights. Identify subtle bugs to avoid unpredictable behavior that could affect users and address security vulnerabilities that threaten your application while gaining knowledge of application security through the Security Hotspots feature. Within moments, you can begin using the platform right where your code resides, benefiting from immediate access to the most current features and updates. Project dashboards provide vital information on code quality and readiness for release, keeping both teams and stakeholders in the loop. Showcase project badges to demonstrate your commitment to excellence within your communities. Code quality and security are essential across your entire technology stack, encompassing both front-end and back-end development. That’s why we support a wide range of 24 programming languages, including Python, Java, C++, and many more. The demand for transparency in coding practices is on the rise, and we invite you to be a part of this movement; it's completely free for open-source projects, making it an accessible opportunity for all developers! Plus, by participating, you contribute to a larger community dedicated to improving software quality.

Eclipse Theia allows you to create multi-language Cloud & Desktop IDEs using state-of-the art web technologies. Are you unsure whether you need a desktop or web version? Theia allows you to create one IDE and have it run in browsers or native desktop applications from a single source. The Eclipse Foundation hosts the Theia project. It is a non-profit corporation and is developed by a diverse group. Projects hosted at an Open-Source Foundation, unlike other open-source projects, are protected from single-vendor decisions that would be detrimental to the interests of the diverse community. Theia is modularly designed to allow users and extenders to customize and extend every aspect of the system. It is easy to create a custom IDE-like product by simply listing all required extensions in a package.json. It is simple to add new functionality by implementing your extensions. This gives you all the flexibility that you need.

StackBlitz WebContainers deliver a browser-based Node.js environment that fundamentally changes what's possible in interactive developer experiences. They allow applications to run native package managers, build tools, and full-stack frameworks fully client-side with no server infrastructure required. Using the WebContainer API, teams can programmatically boot a container, populate its filesystem, install packages, and start dev servers inside the browser. This unlocks advanced use cases like interactive tutorials, live coding courses, in-app IDEs, low-code editors, and AI-powered tooling with runtime awareness. WebContainers work across major browsers, support Wasm out of the box, and provide strong security through sandboxed execution. Developers can offer disposable, fast, and isolated workspaces that scale instantly without backend costs. The technology already powers leading frameworks and education platforms, validating its stability and performance. With WebContainers, any app can embed a complete development environment to increase engagement, accelerate learning, and reduce friction in product workflows.

Typora offers a fluid experience for users, serving effectively as both a writing and reading tool. It eliminates the need for a preview window, mode switcher, or any markdown syntax symbols, thereby cutting out unnecessary distractions. Instead, it features a live preview option that allows you to focus solely on your content. The software also facilitates file management with an easily accessible file tree panel and an article list panel, enabling you to organize your files according to your preferences, including integration with sync services such as Dropbox or iCloud. You can effortlessly view the outline structure of your documents in the outline panel, making navigation throughout the document quick and intuitive, as you can jump to any section with just one click. Additionally, Typora allows you to export your work to PDF complete with bookmarks. You can further enhance your productivity by importing and exporting various other formats, such as docx, OpenOffice, LaTeX, MediaWiki, and Epub. To top it off, you can track your document’s length in terms of words, characters, lines, or estimated reading time, providing you with valuable insights into your writing. This comprehensive functionality makes Typora a powerful tool for anyone involved in writing or content creation.

Visual Studio IntelliCode enhances the development experience through AI-driven assistance. By prioritizing suggestions that you are most likely to need, IntelliCode helps streamline your workflow significantly. Its recommendations are informed by a wealth of information derived from numerous open-source projects on GitHub, each boasting over 100 stars. When used alongside the context of your existing code, the completion suggestions are customized to align with established coding practices. Beyond just completing statements, IntelliCode also offers signature assistance, suggesting the most relevant overload options based on the current context. Not only does IntelliCode provide personalized recommendations for your code, but it also facilitates sharing these insights across your team effortlessly. Thanks to this preview feature, you can create a team model that offers tailored suggestions for code outside the open-source realm, including methods from your custom utility classes or specialized libraries. Furthermore, by integrating our build task into your development pipeline, you can ensure that team completions remain current with any changes in the repository, fostering a more collaborative coding environment. This powerful tool ultimately empowers developers to work more efficiently and effectively within their teams.

GroupBy's headless eCommerce Search & Product Discovery Platform powered by Agent Search on Gemini Enterprise Agent Platform for Retail enhances some of the largest B2B & B2C brands. Built on AI fundamentals, GroupBy's AI-first composable platform is bringing next-generation search technology to retailers & wholesalers worldwide – supplying Google-Quality search results to their online shoppers. The platform consists of Data Enrichment, Search & Recommendations, Merchandising, Analytics & Reporting, providing eCommerce merchants with access to a powerhouse of products & services designed to enhance the digital customer experience. GroupBy platform is transforming eCommerce merchandising from rule-based to revenue-generating, optimizing productivity & efficiencies, & reducing time to market. This allows retailers, wholesalers & distributors to focus on business strategic initiatives that drive revenue. Learn more about how GroupBy is shaping the future of eCommerce by visiting our website and follow us on LinkedIn, Twitter and Instagram.