Search results for OpenRole

SoftDoes is the engineering partner for organizations running mission-critical systems. We deliver results without the fluff — scalable architectures, production-ready AI, and enterprise-grade cloud solutions. With nearly a decade of experience and 350+ projects delivered, our engineers know how to solve complex technical challenges, optimize performance, and accelerate innovation. From custom software development and UI/UX design to managed cloud infrastructure and AI/ML solutions, we build technology that lasts. We stay ahead of the curve with modern toolchains — Infrastructure-as-Code (Terraform, Pulumi), container orchestration (Kubernetes, Docker), CI/CD automation (GitHub Actions, ArgoCD), and observability stacks (Prometheus, Grafana, OpenTelemetry). This ensures transparency, reliability, and measurable outcomes for every client. If you want technology that works at scale — no gimmicks, no excuses — we’re your team. Ready to architect the next phase of your business? Let’s build it.

Zeus-code is an innovative AI coding agent that emphasizes code reuse over generation, offering a more efficient and reliable approach to software development. By vector-searching both private GitLab repositories and public modules, it composes full systems using existing, tested code, thereby reducing bugs and enhancing organizational knowledge with each deployment. Key Features and Functionality: - Fine-Grained Vector Search: Indexes functions, modules, and tests at an appropriate granularity to retrieve snippets that seamlessly fit the intended use. - OpenRouter Cost Optimization: Utilizes the most cost-effective models for each step, including free-tier options, ensuring efficient performance without compromising quality. - Organizational Recipe Memory: Accepted code compositions, along with their diffs and rollbacks, are stored as recipes, facilitating faster and more accurate future development efforts. - Multi-Agent Architecture: Plans to implement specialized agents for planning, designing, engineering, and quality assurance, allowing for a more streamlined and collaborative development process. Primary Value and User Solutions: Zeus-code addresses common challenges in software development by minimizing reliance on large language models (LLMs) for code generation, which often leads to higher costs and increased bug rates. By focusing on reusing existing, battle-tested code, it significantly reduces generation-related bugs, lowers development costs, and accelerates shipping cycles. Additionally, its approach to building a repository of organizational knowledge ensures that development efforts become more efficient over time, as successful compositions are stored and easily accessible for future projects.

Extremis is a shared team inbox and AI Agent for WhatsApp, Instagram, Messenger, and Gmail. A business connects its WhatsApp Business number through Meta's official Cloud API and can keep using the WhatsApp Business app on the same number. It invites its team, and everyone works customer conversations together in real time, with a clear owner for every chat. What teams get: • One inbox for WhatsApp, Instagram, Messenger, and Gmail, including images, voice notes, reactions, and replies • Assignment and routing: round-robin, weighted ratio, or least busy • Open, pending, and closed status, snooze, internal notes, @mentions, quick replies, and labels • Contacts with custom fields, merge, CSV export, and WooCommerce order history • WhatsApp templates and campaigns, sent immediately or scheduled • Reports on response and resolution times, agent performance, and team attendance • An AI Agent that suggests replies or answers customers independently, then hands the chat to a human agent The AI Agent runs on the business's own OpenAI, Anthropic, Google, OpenRouter, or Groq key, and Extremis charges nothing per AI answer. Extremis works in the browser and installs on Android and iPhone with push notifications. Pricing is per workspace, monthly or yearly, with a 7-day free trial and no markup on Meta's WhatsApp fees. Start at https://extremis.online

Coretex SEO is an AI-powered SEO plugin for WordPress, built for site owners, freelancers and agencies who want help with on-page SEO without handing control to an AI. It scans your posts and pages and drafts fixes for page titles, meta descriptions, headings, image titles, image alt text, internal link phrases and spelling. Every suggestion goes into an approval queue, where you can apply it, edit it first or dismiss it. Nothing on the site changes until you approve it. You connect your own AI provider and pay them directly with no markup: OpenAI, Anthropic Claude, Google Gemini, Mistral AI, Groq, DeepSeek, xAI Grok, Perplexity or OpenRouter. You can also run Ollama on your own server, with no API key and no per-request bill. An optional Site Profile (niche, audience, tone, topics) helps the AI write in a way that suits each site. Coretex SEO also includes GEO (Generative Engine Optimisation) tools to help structure content for AI search, such as ChatGPT, Perplexity and Google AI Overviews. You can schedule scans too: the Free Edition has an optional weekly SEO audit with an email report, and the Agency Edition adds Schedule Full Week Scan. Pricing: the Free Edition is free forever and needs no licence key. Agency is £49 per site per year. Requirements: WordPress 6.0 or later and PHP 7.4 or later. First published on WordPress.org in June 2026.

Inferagate is an open-source AI gateway designed for production teams seeking enterprise-grade control over their AI deployments. It serves as a centralized control layer, enabling seamless integration with various inference providers, efficient model management, and comprehensive oversight of AI operations. By offering a unified interface, Inferagate simplifies the complexities associated with managing multiple AI services, ensuring consistent performance and security across applications. Key Features and Functionality: - Unified Provider Integration: Connects with multiple inference providers such as OpenAI, Anthropic, Bedrock, Hugging Face, OpenRouter, vLLM, Ollama, and Groq, allowing teams to manage all AI services through a single gateway. - Model Management: Facilitates the publication of approved models, enabling teams to control and monitor model usage effectively. - Security and Compliance: Implements guardrails, data loss prevention (DLP) measures, and policy enforcement to maintain security standards and compliance requirements. - Operational Oversight: Provides tools for monitoring traffic, investigating requests, and auditing AI operations, ensuring transparency and accountability. - Cost Management: Offers budget controls and financial oversight to manage and optimize AI-related expenditures. Primary Value and Problem Solving: Inferagate addresses the challenges faced by teams managing multiple AI inference providers by offering a centralized, OpenAI-compatible API layer. This consolidation enhances operational efficiency, improves security through consistent policy enforcement, and provides financial control over AI-related costs. By streamlining the integration and management of diverse AI services, Inferagate empowers teams to focus on delivering high-quality AI-driven products without the overhead of complex infrastructure management.

RaptorWrite is an AI-integrated writing tool designed specifically for long-form projects, such as novels. It provides a comprehensive environment where authors can plan, draft, edit, and manage entire manuscripts with AI assistance seamlessly integrated into the workflow. The platform supports cloud-based storage, enabling users to access their work from any device with a web browser, eliminating concerns about data loss due to device failures. Key Features and Functionality: - Multi-Model Integration: RaptorWrite integrates multiple AI providers, including OpenRouter, OpenAI, Anthropic, Google, MistralAI, and DeepSeek, allowing users to switch between models without vendor lock-in. - Model Customization: Users can adjust parameters such as temperature, output length, reasoning effort, and verbosity to tailor the AI's output to their specific needs. - Workflow Tools: The platform offers project management capabilities, prompt organization, model grouping, and user interface customization to streamline the writing process. - Creative Writing Assistance: RaptorWrite provides tools for brainstorming, story outlining, scene expansion, and content generation, aiding authors in overcoming creative bottlenecks. Primary Value and User Solutions: RaptorWrite addresses common challenges faced by fiction writers and content creators, such as character development, plot structuring, and content generation. By integrating multiple AI models and offering extensive customization options, it empowers users to experiment with different AI capabilities without being tied to a single provider. The platform's focus on long-form content and its suite of creative writing tools help authors overcome creative bottlenecks, streamline their writing process, and meet tight publication deadlines.

PDF Pals is a native macOS application designed to revolutionize the way users interact with PDF documents. By integrating advanced AI capabilities, it allows users to engage in intuitive conversations with their PDFs, extracting key information and insights without the need to manually sift through extensive content. This tool is particularly beneficial for professionals who regularly work with large volumes of documents, such as researchers, legal experts, and financial analysts. Key Features and Functionality: - AI-Powered Chat: Engage in interactive dialogues with your PDF documents to quickly obtain relevant information and summaries. - Optical Character Recognition (OCR): Effectively processes scanned PDFs and complex forms, making them searchable and interactive. - Local Data Processing: Ensures privacy by processing all documents locally on your Mac, without uploading to external servers. - Flexible API Integration: Supports multiple API providers, including OpenAI, Azure OpenAI Service, and OpenRouter, allowing users to choose their preferred service. - No File Size Limitations: Handles PDFs of any size or complexity without restrictions. - Multi-PDF Handling: Import and interact with multiple related PDFs simultaneously for comprehensive analysis. - Customizable Tagging: Offers tagging features for analyzed PDFs, which can be customized for better organization. - One-Time Payment Model: Provides a cost-effective solution with a one-time payment, avoiding recurring subscription fees. Primary Value and User Solutions: PDF Pals addresses the common challenge of efficiently extracting and analyzing information from extensive PDF documents. By enabling AI-driven interactions, it significantly reduces the time and effort required to review, summarize, and comprehend complex materials. This enhancement in productivity is invaluable for professionals who need to process large volumes of information accurately and swiftly. Additionally, its commitment to privacy through local data processing ensures that sensitive documents remain secure, making it a trustworthy tool for handling confidential information.

RubyLLM is a streamlined Ruby API designed to unify interactions with various AI models, including GPT, Claude, Gemini, and more. It simplifies the development of chatbots, AI agents, retrieval-augmented generation (RAG) applications, and content generators by providing a consistent interface across different AI providers. With minimal dependencies—only Faraday, Zeitwerk, and Marcel—RubyLLM ensures lightweight integration into Ruby applications. Key Features and Functionality: - Conversational AI: Facilitates seamless chatbot development using `RubyLLM.chat`. - Vision Capabilities: Enables analysis of images and videos. - Audio Processing: Supports transcription and understanding of speech through `RubyLLM.transcribe`. - Document Handling: Extracts information from various file types, including PDFs, CSVs, and JSON. - Image Generation: Creates images with `RubyLLM.paint`. - Embeddings: Generates embeddings using `RubyLLM.embed`. - Content Moderation: Ensures content safety with `RubyLLM.moderate`. - Tool Integration: Allows AI to call Ruby methods, enhancing functionality. - Structured Output: Provides JSON schemas for organized data handling. - Streaming Responses: Delivers real-time responses using blocks. - Rails Integration: Offers ActiveRecord integration with `acts_as_chat`. - Asynchronous Processing: Utilizes fiber-based concurrency for efficient operations. - Extensive Model Support: Maintains a registry of over 500 models with capability detection and pricing. - Multiple Providers: Supports various AI providers, including OpenAI, Anthropic, Gemini, VertexAI, Bedrock, DeepSeek, Mistral, Ollama, OpenRouter, Perplexity, GPUStack, and any OpenAI-compatible API. Primary Value and User Solutions: RubyLLM addresses the complexity of interacting with multiple AI providers, each with its own client libraries, APIs, and response formats. By offering a unified and elegant API, it eliminates the need to manage disparate clients, streamlining the development process. This consistency allows developers to focus on building robust AI-powered applications without the overhead of integrating and maintaining multiple provider-specific clients. Whether developing chatbots, AI agents, or content generators, RubyLLM provides the tools necessary to build sophisticated applications efficiently.

Discover Trusys.ai, the first full-stack AI assurance platform that lets you evaluate, red-team, and continuously monitor your GenAI applications—across text, image, audio, and video. Whether you're launching a chatbot, deploying RAG pipelines, building agentic systems, or managing enterprise-scale AI deployments—Trusys helps you stay compliant, secure, and performant from day one. ✅ Evaluate accuracy, bias, hallucination & fairness ✅ Run security testing mapped to MITRE ATT&CK, OWASP-LLM, EU AI Act ✅ Monitor live AI behavior with Tru Pulse: sampling, OpenTelemetry trace integration, real-time alerts ✅ Build or import datasets (text, audio, etc.), with full PII encryption and audit logs ✅ No-code interface + Promptfoo-powered evaluation engine 🔒 Works for BFSI, healthcare, SaaS, and government AI teams

VerbaGPT is an innovative tool designed to simplify data analysis by enabling users to interact with their data using natural language queries. By leveraging large language models (LLMs), VerbaGPT allows users to ask questions about their SQL databases, CSV files, or personal notes and receive immediate, insightful answers. The platform emphasizes data privacy by operating locally on the user's hardware, ensuring that sensitive information remains secure. Key Features and Functionality: - Advanced Analytics: Beyond basic data aggregation, VerbaGPT supports complex queries, data modeling, and the creation of visualizations, enabling comprehensive data analysis. - Text-to-Python Capability: The platform translates natural language queries into Python code, encompassing SQL operations and a wide range of statistical and visualization functions. - Data Privacy: VerbaGPT runs locally, ensuring that the LLM does not have direct access to the user's data. Only schema information explicitly provided by the user is shared, maintaining data confidentiality. - User-Friendly Design: The application is intuitive, allowing users to upload data in CSV/TXT formats, connect to various SQL databases, and select from curated models, including free options. - Offline Functionality: VerbaGPT offers an experimental offline mode, enabling users to perform data analysis without an internet connection. - Error Recovery and Analysis: Features like the 'retry' button help recover from code execution errors, while the 'analyze' button provides commentary on code and results. - Model Training and Management: Users can train models with successful question-SQL pairs, flag unsuccessful generations, and toggle between different LLMs such as OpenRouter and OpenAI. Primary Value and User Solutions: VerbaGPT addresses the challenge many organizations face in becoming data-driven by bridging the gap between data accessibility and technical expertise. It empowers non-technical users to extract meaningful insights from their data without requiring coding skills, thereby democratizing data analysis. For technical users, VerbaGPT enhances productivity by streamlining the data querying process. By prioritizing data privacy and offering a user-friendly interface, VerbaGPT ensures that users can confidently and efficiently interact with their data to inform decision-making and drive innovation.

Aegitox is an enterprise-grade AI moderation infrastructure for Discord communities. It replaces manual oversight with a deterministic AI pipeline, deploying a zero-latency defensive grid that intercepts, deletes, and replaces toxic payloads before any community member can read them. CORE AI INTERCEPTOR & ARCHITECTURE Built on strict constant-time execution mandates with zero-allocation memory management, Aegitox guarantees algorithmic supremacy with 2-12ms response times. Toxicity is intercepted before it renders on the client side. The semantic pipeline is powered by a dual MiniLM-L6-v2 neural network matrix. Gate 1 detects toxicity by calculating a normalized semantic confidence vector (0.0 to 1.0), isolating genuine hostility from friendly banter. Gate 2 calculates the exact directional threat vector (Personnel, Self, or System). Based on the target, threats are mapped to a pool of 120 static, pre-vetted de-escalation placeholders to instantly neutralize drama without relying on generative AI hallucinations. THE END OF LEGACY MODERATION Legacy bots delete messages after the chat reads them, log chat history to unsecured databases, and burn out volunteer staff. Aegitox solves this by surgical, target-aware de-escalation that softens the atmosphere without muting the entire chat. AUTOMATED PENALTY ENGINE Aegitox tracks behavioral patterns using a detailed karma metric. Breaching custom karma thresholds triggers autonomous disciplinary action. In Judicial Mode, the bot generates a private Discord Thread for appeals, ensuring staff retain the final verdict. If an appeal is rejected, the system autonomously doubles the Time Out duration penalty to deter trolls. For massive guilds facing bot raids, Zero-Tolerance Mode instantly executes Discord's native cryptographic gag, bypassing appeals entirely. ENTERPRISE PRO INFRASTRUCTURE Complex moderation routing is handled via a VIP priority lane directly to our dedicated OpenRouter Deep AI endpoints. Administrators have total system prompt governance, allowing them to inject custom instructions to dictate the bot's exact tone and personality. DATA PRIVACY & ZERO-TRUST SECURITY Aegitox enforces absolute data sovereignty through a strict Zero Data Retention (ZDR) architecture. All text processing occurs exclusively in volatile RAM. Data is mathematically purged the exact millisecond the threat evaluation completes. Chat logs are never written to a database, never logged to disk, and explicitly never utilized to train AI models. Secure enterprise billing is handled exclusively by our authorized Merchant of Record, PayPro Global.

ScriptFlo is web-based screenwriting software for writers who want one app for the whole process: planning, writing, coaching and production prep. It runs in the browser on desktop and mobile, saves to the cloud with version history, and keeps every feature in a single plan. Write. The editor applies industry-standard screenplay formatting as you type. Typing INT. or EXT. starts a scene heading, Tab and Enter move through element types, and margins stay locked to the standard, so you never format by hand. Distraction-free mode, full undo history and auto-save are included. Plan. Nine story structure templates come pre-loaded with beat descriptions and page targets: Three-Act, Save the Cat, the Hero's Journey, Dan Harmon's Story Circle, Freytag's Pyramid, the Sequence Method and three TV formats, plus custom templates you define. Corkboard, canvas, storyboard and timeline views sync both ways with the script, so reordering scenes visually updates the draft. Coach. The AI assistant reads your entire script, not just the current scene. It gives scene notes, checks character voice, proposes alternate directions and punches up dialogue on request. It never writes over your work. Bring your own API key for Claude, OpenAI or OpenRouter models; ScriptFlo does not mark up AI usage. Produce. Tag scene elements across 18 breakdown categories, from cast and props to stunts and VFX, and export breakdown reports for a line producer. Scenes are numbered automatically. Export and keep your work. PDF with title page, Final Draft (.fdx), Fountain, HTML and plain text. Import existing scripts from Fountain or FDX files. If you cancel, you keep access to open and save the scripts you created. Pricing. One plan at $30 per month with everything included: unlimited projects, all templates, AI coaching, every export format, planning views and production breakdown. Every account starts with a 14-day free trial (a card is required, nothing is charged until the trial ends, cancel any time). Founding members lock in their price for life. Built for first-time screenwriters who want structure and formatting handled for them, and for working writers moving from Final Draft, WriterDuet, Celtx or Arc Studio who want an all-in-one tool with a writing coach included.

Rule Learner is a machine learning tool designed to help business analysts extract classification rules from historical data without requiring extensive programming or data science expertise. By inputting historical data into two simple Excel tables, users can automatically generate decision tables that classify new data records based on past classifications. The generated rules are presented in a user-friendly Excel format, allowing for easy download and further analysis. Key Features and Functionality: - Automated Rule Generation: Utilizes machine learning algorithms to derive classification rules from historical data. - Excel Integration: Accepts input data in Excel tables and outputs the generated rules in an Excel decision table format. - Algorithm Selection: Offers the flexibility to choose from different machine learning algorithms to optimize rule accuracy. - User-Friendly Interface: Designed for business analysts, eliminating the need for advanced programming or data science skills. Primary Value and Problem Solved: Rule Learner empowers business analysts to harness the power of machine learning for data classification tasks without the steep learning curve typically associated with data science. By simplifying the process of rule extraction and providing results in familiar Excel formats, it enables organizations to make data-driven decisions more efficiently and accurately.

OpenCover makes it easy and affordable to protect your portfolio against onchain risks. Get cover against smart contract hacks, oracle failures and more from vetted underwriters. On Base, Optimism and other leading blockchain networks.

High-quality, ubiquitous, and portable telemetry to enable effective observability

OpenModerator offers cutting-edge, open-source AI tools for content moderation.

OpenNovel is an advanced AI-powered platform designed to provide smooth, human-like translations of web novels, enabling readers to access content across languages effortlessly. By leveraging cutting-edge AI models such as ChatGPT, Gemini, and DeepSeek, OpenNovel ensures high-quality translations without content filters or word limits. Users can translate websites, text, and ePub files, maintaining consistency in names, genders, and terms through a customizable glossary. With features like auto-translation and ePub export, OpenNovel caters to both casual readers and dedicated novel enthusiasts. Key Features and Functionality: - AI-Powered Translation: Utilizes advanced AI models to deliver fluent and natural translations of web novels. - Versatile Input Options: Supports translation of content from websites, text, and ePub files. - Customizable Glossary: Allows users to define and maintain consistency in character names, genders, and specific terms throughout the novel. - Auto-Translation and Export: Enables automatic translation of entire novels and export to ePub format for offline reading on e-readers. - Bulk Import: Facilitates the import of multiple chapters from supported websites, streamlining the translation process. - Editing Capabilities: Provides tools to edit translations directly, ensuring accuracy and personalization. Primary Value and User Solutions: OpenNovel addresses common challenges faced by readers of machine-translated (MTL) websites and self-translators, such as slow release schedules, inconsistent translation quality, intrusive advertisements, and cumbersome content management. By offering a seamless and efficient translation experience, OpenNovel empowers users to access a vast array of novels in their preferred language, enhancing their reading experience and expanding their literary horizons.

OpenHome is an AI-powered voice interaction platform that enables developers to transform any device into a smart, voice-enabled system. By leveraging OpenHome's Voice SDK, users can create natural, conversational experiences across various hardware devices and applications, including smart home technologies, IoT devices, and enterprise solutions. The platform emphasizes community-driven development, offering tools that are both cutting-edge and accessible, allowing developers of all levels to contribute to a versatile and customizable Voice AI ecosystem. Key Features and Functionality: - Customizable AI Personalities: Developers can create and manage AI voice agents, known as Personalities, tailored for specific tasks and applications. Each Personality can be customized in terms of description, purpose, voice, and behavior, allowing for deeply personalized voice interactions. - Extensible Abilities: The platform supports the addition of Abilities, which are plugins that enhance the functionality of Personalities. These Abilities can integrate third-party APIs, add new skills, or perform specific tasks, thereby expanding the capabilities of the AI agents. - Marketplace for Sharing and Discovery: OpenHome offers a Marketplace where users can upload, share, and discover Personalities and Abilities created by the community. This fosters collaboration and innovation, enabling developers to build upon each other's work and accelerate development. - Comprehensive Dashboard: The web-based dashboard allows users to view, create, and modify Personalities and Abilities, manage conversations, review conversational history, and configure system settings. It serves as a central hub for managing all aspects of the OpenHome ecosystem. - Developer-Friendly APIs: OpenHome provides powerful APIs for speech-to-text, text-to-speech, and language understanding, enabling developers to integrate advanced voice interaction capabilities into their applications seamlessly. Primary Value and User Solutions: OpenHome addresses the growing demand for natural and intuitive voice interactions by providing a platform that is both powerful and accessible. It empowers developers to create AI voice agents that can understand and respond to user inputs in a human-like manner, enhancing user engagement and satisfaction. By offering customizable Personalities and extensible Abilities, OpenHome allows for tailored solutions across various industries, from smart home automation to enterprise applications. The community-driven Marketplace further accelerates development by enabling the sharing and discovery of new functionalities, fostering a collaborative environment that drives innovation in voice AI technology.

The AI Agent Infrastructure Startups Landscape report, updated in May 2026, provides a comprehensive overview of the leading startups developing horizontal infrastructure for AI agents. This landscape is categorized into three primary sub-sectors: durable execution and workflow orchestration, observability and evaluation, and gateway/proxy services. Key Features and Functionality: 1. Durable Execution and Workflow Orchestration: - Temporal: Secured approximately $430 million in cumulative funding, with a $300 million Series D round in February 2026, valuing the company at $5 billion. Notable for 380% year-over-year revenue growth and over 20 million monthly installs. - Inngest: Raised around $34 million cumulatively, including a $20.5 million round in May 2025. Introduced Temporal-compatible workflows in February 2026. - Restate: Obtained $7 million in seed funding, launching commercially in March 2026 with a focus on lightweight workflows-as-code. 2. Observability and Evaluation: - Braintrust: Accumulated approximately $120 million in funding, with an $80 million Series B round, valuing the company at $800 million. Serves clients like Notion, Replit, Cloudflare, Ramp, and Dropbox. - LangSmith (LangChain Inc): Part of LangChain Inc's cumulative funding of over $125 million, offering native observability for LangChain and LangGraph stacks. - Helicone: Secured around $10 million in funding, operating as an open-source AI gateway and observability tool, and is a Y Combinator alumnus. - Langfuse: Acquired by ClickHouse in January 2026 as part of ClickHouse's $400 million Series D at a $15 billion valuation; continues to maintain its open-source code and active community. - AgentOps: Raised approximately $2.6 million in seed funding, partnered with Red Hat, and backed by 645 Ventures, Afore Capital, and Plug and Play. - Arize Phoenix: Part of Arize AI's cumulative funding exceeding $100 million, offering open-source observability within Arize's broader MLOps platform. 3. Gateway and Proxy Services: - LiteLLM (BerriAI): Raised around $15 million across seed and Series A rounds, achieving approximately 418 million monthly PyPI downloads as an OpenAI-compatible multi-vendor proxy. - Portkey: Secured about $10 million in funding, functioning as a gateway and observability hybrid with a strong presence in the Asia-Pacific region. - OpenRouter: Bootstrapped with undisclosed funding, specializing in multi-model aggregation and operating profitably. Primary Value and Solutions Provided: These startups are instrumental in building the foundational infrastructure necessary for the efficient operation of AI agents. By offering durable execution environments, comprehensive observability tools, and robust gateway services, they address critical challenges in AI agent deployment, such as ensuring reliable execution, monitoring performance, and facilitating seamless integration across various platforms. Their innovations enable businesses to deploy AI agents more effectively, leading to enhanced automation, improved decision-making processes, and streamlined operations.

Agents can act. They don't know who. Every agentic workflow eventually hits the same wall: who needs to approve this, get notified, or pick up the work next? Hard-coding an org chart into a prompt is brittle — it breaks the moment someone's on leave, a limit changes, or the org restructures. Roster fixes that with one primitive: resolve. Ask in plain language — "Who should approve the Helios renewal?" — over MCP, CLI, or REST, and Roster matches the right participant (a business role, not a named person) using labels, typed metadata, and approval limits, then expands to the current live human or agent through your directory and any active delegations. Someone out on leave? Roster returns their delegate automatically, with full provenance, and never rewrites your directory to do it. It ships as a single Docker container — no separate database, no extra infrastructure — exposing four surfaces from one runtime: an MCP server (the primary agent surface), REST API, CLI, and a web dashboard for admins and project owners. Connects to Microsoft Entra ID, Okta, Google Workspace, Workday, LDAP/Active Directory, or CSV. Bring your own LLM (OpenAI, Anthropic, or Mistral direct, or route through OpenRouter as a gateway) — Roster never marks up token costs. Every resolution is logged as an append-only audit event; every model call is observable by provider, tokens, and cost; PII is redactable field by field. Built by Advantys, the team behind WorkflowGen's routing and delegation engine — in production for 20+ years across 500+ organizations. Key capabilities - Natural-language resolve queries over MCP, CLI, or REST — get back the human or agent who should act, in real time - Participants model business responsibility (roles, approval limits, regions), not named people — routing survives org change - Delegations apply as runtime overlays for absences and handovers, with full provenance, without touching upstream directory data - Six directory connectors — Microsoft Entra ID, Okta, Google Workspace, Workday, LDAP/Active Directory, CSV — with cache-only, cron-scheduled sync - Three first-class identity types: team members, AI agents, and service accounts - Append-only audit log and per-run model observability (provider, tokens, cost) with field-level PII redaction - Draft → test → live → archived project lifecycle, so new routing logic can be tested without polluting production history - Single Docker container (advantys/roster), self-hosted on every tier, bring-your-own LLM - Five SSO options: Microsoft Entra ID, Okta, Google, generic OAuth, SAML