Atomic Agent is an open-source, local-first AI agent runtime. It runs on your own machine on top of llama.cpp, drives your desktop — browser automation, file editing, shell commands — with dangerous actions gated behind an approval step, and keeps sessions, memory, and traces on your own disk. It works with local models as well as OpenAI-compatible and OpenRouter providers. CLI and TUI, macOS and Linux, with a Windows developer preview. MIT-licensed.
GitHub: https://github.com/AtomicBot-ai/atomic-agent
Site and docs: https://atomicagent.io
Search for images and receive available image URLs, thumbnails, dimensions and source-page links as structured data. Use it for image discovery and source attribution. This API returns search data; it does not download or license the images.
Use the existing REST API, MCP or CLI client. API usage requires a FlashData account, API key and paid Credits. The MIT-licensed client is free to install; API data usage is not a free API. Follow the linked source documentation for the supported inputs, response fields, authentication and current pricing.
Documentation: https://flashdata.dev/docs/sources/google_images
Search a topic and region to retrieve Google News results with titles, source links, publication details and available snippets. Build topic-monitoring or news-discovery workflows. Results are search data, not full news articles.
Use the existing REST API, MCP or CLI client. API usage requires a FlashData account, API key and paid Credits. The MIT-licensed client is free to install; API data usage is not a free API. Follow the linked source documentation for the supported inputs, response fields, authentication and current pricing.
Documentation: https://flashdata.dev/docs/sources/google_news
Search a place or local-business query to retrieve structured place results, addresses and available rating or contact fields. Use it for local discovery and business research. Contact details may be missing; this is not a complete business database.
Use the existing REST API, MCP or CLI client. API usage requires a FlashData account, API key and paid Credits. The MIT-licensed client is free to install; API data usage is not a free API. Follow the linked source documentation for the supported inputs, response fields, authentication and current pricing.
Documentation: https://flashdata.dev/docs/sources/google_places
Search a product query to retrieve a Google Shopping result snapshot with product, merchant and available price fields. Use the structured snapshot for product research. It does not guarantee inventory, final purchase prices or continuous monitoring.
Use the existing REST API, MCP or CLI client. API usage requires a FlashData account, API key and paid Credits. The MIT-licensed client is free to install; API data usage is not a free API. Follow the linked source documentation for the supported inputs, response fields, authentication and current pricing.
Documentation: https://flashdata.dev/docs/sources/google_shopping
MLP Software Development GmbH is a premier software engineering firm based in Osnabrück, Germany. We specialize in building custom mobile applications (iOS & Android) and scalable web platforms for the "Mittelstand" (mid-sized companies) and large enterprises.
We move beyond standard solutions to deliver software that solves complex business challenges: from workflow automation to seamless ERP/SAP integration. Our team ensures 100% GDPR compliance and high-security standards.
Core Competencies:
Native Mobile Apps: High-performance iOS (Swift) and Android (Kotlin) development.
Enterprise Web Platforms: Complex B2B portals and internal tools.
System Integration: Seamless connection with SAP, Microsoft, and Legacy systems via API.
Cloud & DevOps: Scalable infrastructure on AWS/Azure.
100% Made in Germany. Engineering excellence you can trust.
greenable's Software ermöglicht es Unternehmen, den CO₂-Fußabdruck ihrer Produkte – den sogenannten Product Carbon Footprint (PCF) – effizient und automatisiert zu berechnen. Die Lösung ist zertifiziert nach DIN EN ISO 14067 und GHG-Protokoll und richtet sich speziell an den produzierenden Mittelstand in der Automobil-, Stahl- und Kunststoffbranche.
Dank der intuitiven Benutzeroberfläche und automatisierten Datenerfassung sparen Unternehmen wertvolle Ressourcen. Stücklisten und Arbeitspläne aus ERP-Systemen sowie EMS-Daten für präzise Energiewerte können per Upload importiert werden. Der Product Carbon Footprint Monitor von greenable nutzt Emissionsfaktoren etablierter Datenbanken wie ecoinvent und ermöglicht das Einpflegen von lieferantenspezifischen Daten. Die Ergebnisse sind sofort kommunizierbar und können mit einem Klick als Kurzbericht ausgeleitet werden.
trønso ist eine Assessment-Software für HR-Teams in Deutschland. Statt Lebensläufe manuell zu sichten, legen kleine und mittlere Unternehmen selbst fest, welche Fähigkeiten eine Stelle verlangt, und bilden diese in einem eigenen Online-Test ab. Zur Verfügung stehen neun Fragetypen, darunter Multiple Choice, Kontrollkästchen, Freitext, lineare Skalen, Raster und Sortierung; ausführbare Code-Aufgaben werden auf Anfrage freigeschaltet. Bewerber erhalten einen zeitlich begrenzten Einladungslink per E-Mail. Nach der Abgabe stehen die Auswertung mit Punkten pro Frage und ein Vergleich aller Kandidaten sofort bereit. Zeitlimit, Aktivitätsprotokoll und optionales Kamera-Proctoring sorgen für vergleichbare Ergebnisse. Alle Daten liegen auf Servern in der EU, ein Auftragsverarbeitungsvertrag steht automatisch bereit. trønso wird in Deutschland entwickelt und betrieben.
Axerto is an Advertisement Conversion Engine designed for performance marketers and PPC agencies. The platform integrates directly with the Google Ads API to synchronize campaign data -including high-performing headlines, sitelinks, and keywords - to automatically generate exact-match landing pages. This process is engineered to improve Ad Relevance and Quality Scores by strictly aligning the ad copy with the destination page content.
Key technical capabilities include automated conversion tracking (auto-configuring AW- tags, GA4, and Google Tag Manager), native iOS click attribution (capturing gclid, gbraid, and wbraid parameters), and built-in Enhanced Conversions using SHA-256 data hashing.
Additionally, Axerto provides a centralized dashboard for lead and client management, custom domain routing with automatic SSL provisioning, real-time webhook integrations for external CRMs, and the automated generation of page-specific legal documents (Privacy Policy and Terms of Service) to maintain ad network compliance.
Percol ist die souveräne Automatisierungsplattform für den Mittelstand. Wir binden die Branchensoftware der Betriebe an, etwa Auftragsverwaltung, Plantafel oder Buchhaltung, und automatisieren die Arbeit dazwischen.
Ein Beispiel: Percol generiert aus eingehenden Support anfragen automatisch Tickets mit allen auffindbaren Informationen. KI-Agenten übernehmen die Routine, wichtige Schritte gibt ein Mensch frei.
Betriebe bekommen bei uns fertige Prozesse statt eines weiteren Tools: Wir bauen, betreiben und überwachen die Automatisierung. Los geht es mit einem Prozess, weitere folgen Schritt für Schritt, ganz ohne Systemwechsel. Für Branchensoftware-Anbieter heißt das: Sie bringen KI zu ihren Kunden, ohne selbst zu entwickeln.
Das Ganze läuft auf europäischer Infrastruktur mit Open-Weight-Modellen statt US-KI-Anbietern.
OpenScheduler is WonderCal’s MIT-licensed, off-the-shelf scheduling toolkit for people building AI scheduling workflows. It provides ready-to-use MCP, REST, Claude-agent, and webhook components for syncing Google Calendar, Microsoft Outlook, and Apple Calendar. WonderCal’s hosted service adds multi-person availability, group scheduling, required and optional participants, cross-company coordination, fallback availability collection, and completed booking across calendar domains. Together, they let AI assistants coordinate and finish complex meetings without sending participants a booking link or exposing private calendar details.
Provide a YouTube video ID to retrieve structured video metadata and available chapters or format information. Use it to inspect a video before building a research workflow. Some fields may be missing; this API does not itself download media.
Use the existing REST API, MCP or CLI client. API usage requires a FlashData account, API key and paid Credits. The MIT-licensed client is free to install; API data usage is not a free API. Follow the linked source documentation for the supported inputs, response fields, authentication and current pricing.
Documentation: https://flashdata.dev/docs/sources/youtube_metadata
Provide a video ID and language to retrieve available transcript segments with timestamps and fullText. The default source is automatic captions when available. It does not generate speech recognition, translate subtitles or fall back when the required track is missing.
Use the existing REST API, MCP or CLI client. API usage requires a FlashData account, API key and paid Credits. The MIT-licensed client is free to install; API data usage is not a free API. Follow the linked source documentation for the supported inputs, response fields, authentication and current pricing.
Documentation: https://flashdata.dev/docs/sources/youtube_transcript
Send a video-search query to retrieve Google video results across websites. Use it to discover sources beyond YouTube and inspect the available result fields. This is Google video search, separate from searching within YouTube.
Use the existing REST API, MCP or CLI client. API usage requires a FlashData account, API key and paid Credits. The MIT-licensed client is free to install; API data usage is not a free API. Follow the linked source documentation for the supported inputs, response fields, authentication and current pricing.
Documentation: https://flashdata.dev/docs/sources/google_videos
Provide a video ID and language to retrieve available uploader-caption segments and fullText. This differs from the transcript capability, which defaults to automatic captions. It does not generate captions, translate tracks or automatically fall back to another source.
Use the existing REST API, MCP or CLI client. API usage requires a FlashData account, API key and paid Credits. The MIT-licensed client is free to install; API data usage is not a free API. Follow the linked source documentation for the supported inputs, response fields, authentication and current pricing.
Documentation: https://flashdata.dev/docs/sources/youtube_captions
Search YouTube by keyword and retrieve structured video results for discovery workflows. The max_results input ranges from 1 to 20; actual results may be fewer. For larger requested result sets, use the separate Extended Search capability.
Use the existing REST API, MCP or CLI client. API usage requires a FlashData account, API key and paid Credits. The MIT-licensed client is free to install; API data usage is not a free API. Follow the linked source documentation for the supported inputs, response fields, authentication and current pricing.
Documentation: https://flashdata.dev/docs/sources/youtube_search
Provide a YouTube video ID to inspect the available license metadata for a Creative Commons label. Use it as a metadata signal in a research workflow. It does not read creator third-party training settings or grant permission to train on, download or reuse content.
Use the existing REST API, MCP or CLI client. API usage requires a FlashData account, API key and paid Credits. The MIT-licensed client is free to install; API data usage is not a free API. Follow the linked source documentation for the supported inputs, response fields, authentication and current pricing.
Documentation: https://flashdata.dev/docs/sources/youtube_trainability
Provide a YouTube channel handle to retrieve channel information and recent videos. The video_limit input ranges from 1 to 100; available statistics and other fields may be incomplete. Use it for a channel snapshot rather than assuming a complete historical archive.
Use the existing REST API, MCP or CLI client. API usage requires a FlashData account, API key and paid Credits. The MIT-licensed client is free to install; API data usage is not a free API. Follow the linked source documentation for the supported inputs, response fields, authentication and current pricing.
Documentation: https://flashdata.dev/docs/sources/youtube_channel
Toolscase AI at https://ai.toolscase.com is a task-based AI model recommender, cost estimator and execution router built by Devonsite, part of Toolscase ecosystem (35+ tools at toolscase.com, Voice Studio at voice.toolscase.com, 108 PDF Tools at pdf.toolscase.com).
Problem: Leaderboards are useless. Best model depends on task - complex reasoning vs long context vs coding vs vision vs JSON formatting vs tool calling. Users waste $200+/mo testing same prompt across models with 77x cost difference from $20.00/1M (Claude Opus 4.5) to $0.26/1M (Llama 3.3 70B).
Solution: Describe what you're trying to accomplish in plain English: "Debug my Laravel application", "Analyze a 300-page PDF", "Write an SEO article", "Build a Chrome extension". Our homepage asks "Which AI should you use? Describe what you're trying to accomplish and we'll recommend the AI model best matched to the job — weighed on quality, cost, speed and capability." We analyze Task Demand across reasoning, long context, coding, vision, JSON, tool calling, then score and recommend best from 12 models across 7 providers.
12 models · 7 providers we route across (from providers page screenshot):
Anthropic 3 models - Maker of Claude family, known for long-context reasoning and coding: Claude Opus 4.5 200K $20, Sonnet 4.5 $12, Haiku 4.5 $4
OpenAI 3 models - Maker of GPT family: GPT-5.1 400K $7.81, GPT-5.1 mini $1.56, GPT-4o mini $0.49
Google 2 models - Maker of Gemini family, very large context windows: Gemini 3 Pro 1M $9.50, Gemini 2.5 Flash 1M $1.95
DeepSeek 1 - Strong reasoning and coding at low cost: DeepSeek V3.2 164K $0.39
xAI 1 - Maker of Grok family: Grok 4 256K $12.00
Meta 1 - Open-weight Llama widely hosted inexpensive: Llama 3.3 70B 131K $0.26
Mistral AI 1 - European strong price/performance: Mistral Large 128K $5.00
Our providers page states: "We route across these providers. Which one answers your task is a consequence of the task, not a preference — the router picks the model, and the model determines the provider."
Key Features: Task Demand Analysis, Optimization & Budget Controls (Best value / Quality / Speed / Cost presets, custom budget ceilings, pre-execution cost estimates, exact post-execution token spend), Automatic Failover Routing (reroutes to next best within budget if primary fails), Interactive Model Explorer & Task Guides.
Pricing: Guest 3 free recommendations/day no signup. Free account unlocks execution, query history, routing controls. Pay-per-use $0.26-$20/1M blended. Alternative to OpenRouter, Portkey, LiteLLM, Vercel AI Gateway, Helicone, Langfuse.
Built for developers, teams and businesses tired of guessing which AI to use.
InnReply is an AI chat box for the websites of small inns, bed and breakfasts, motels and cabin resorts. It answers guest questions using only the property's own web pages, such as check-in times, policies, room details, breakfast and pet rules, and passes any question it cannot answer to the innkeeper by email, where a reply goes straight to the guest. When a guest asks about dates, it checks which rooms are free and opens the property's own booking page with the dates already filled in, so bookings and payments stay with the property. With supported booking systems, it also shows each free room with a photo and price. It matches the website's colors, sends a weekly summary of guest questions, and works on common website builders with a one-line install or an installation done by the InnReply team. A Free plan covers one location and 40 chats a month, and a paid plan covers up to two locations and 1,000 chats a month.