home.social

#a2a — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #a2a, aggregated by home.social.

  1. 30 секунд вместо 30 минут: как мы автоматизировали генерирование конфигураций потоковой обработки с помощью RAG и A2A

    Привет, Хабр! Меня зовут Дмитрий Титов, я DevOps-инженер в команде интеграционных сервисов Platform V Synapse в СберТехе. Наша команда работает над продуктом Platform V Streaming Event Processing — программным решением для фильтрации и трансформации форматов событий, агрегирования и выявления аномалий и закономерностей. В этой статье я расскажу, как мы создали систему автоматического генерирования конфигураций для одного из компонентов нашего продукта, используя RAG (Retrieval-Augmented Generation), векторные базы данных и межагентное взаимодействие по протоколу A2A.

    habr.com/ru/companies/sberbank

    #сбертех #synapse #platform_v #platform_v_synapse #rag #a2a #автоматизация #dsl

  2. Claude Code + EClawbot: assign a Kanban card to your AI coding agent, and it autonomously reads the task, edits code, runs tests, opens a GitHub PR, and moves the card to Done.

    Built-in permission model: pre-authorize safe ops, block dangerous ones, prompt for everything else.

    Setup in 5 minutes: github.com/HankHuang0516/EClaw

    #AI #ClaudeCode #DevTools #A2A #OpenSource

  3. A2A Protocol v0.3 刚发布,三个重要更新:gRPC 双向流支持(低延迟高吞吐)、签名 Agent Card(解决信任问题)、延迟感知路由(Twilio 已用于语音通话自动路由)。

    有意思的是,EClaw 的多 Agent 架构早已实现了类似模式。Entity Identity API 对应 Agent Card 的自描述能力,speak-to API 对应 A2A 的任务委派,Kanban Board 则实现了 A2A 还在设计中的结构化任务生命周期管理。

    A2A 负责 Agent 间通信,MCP 负责 Agent 连接工具,两者互补而非竞争。

    cloud.google.com/blog/products

    a2a-protocol.org/latest/

    #A2A #MCP #OpenClaw #AI #MultiAgent

  4. The AI Agent Internet: A New Infrastructure Layer for Autonomous Digital Systems
    Exploring the OwnAether network’s AI Infrastructure, Capability Bus, AI agent economy, Blockchain,AI Ad Network, and next-generation digital marketplaces…

    medium.com/@ownaether/the-ai-a

    #OwnAether #AI #AIInfrastructure #AIAgents #PersonalAI #AIOS #AIArchitecture #AIFramework #MCP #FutureTech #Web3 #Crypto #Blockchain #EnterpriseAI #AIAutomation #A2A #ACP #AGI #API #FutureOfWork #RewardsNetwork #Startup #PreLaunch

  5. Quite an interesting forward-looking conversation on this week's #ITOps Query vodcast!

    Outshift by Cisco published a whitepaper titled "Scaling out Superintelligence" in January that suggests that new standard network layers will be required for #AIagents and humans to work together at scale. It also began designing a distributed infrastructure to support those layers, which it calls the "#InternetofCognition," that helps agents and humans think together with a shared semantic context. In this episode, Guillaume De Saint Marc, vice president of engineering at Outshift, discusses how the project evolved and what's coming next.

    In today’s episode, we’ll cover…

    -- #Artificialsuperintelligence – definition and challenges

    -- Proposed #OSI model Layers 8 and 9 for #agenticAI

    -- The Internet of Cognition vs #Agntcy and #A2A protocols

    And more!

    lnkd.in/epkprGp4

  6. Quite an interesting forward-looking conversation on this week's #ITOps Query vodcast!

    Outshift by Cisco published a whitepaper titled "Scaling out Superintelligence" in January that suggests that new standard network layers will be required for #AIagents and humans to work together at scale. It also began designing a distributed infrastructure to support those layers, which it calls the "#InternetofCognition," that helps agents and humans think together with a shared semantic context. In this episode, Guillaume De Saint Marc, vice president of engineering at Outshift, discusses how the project evolved and what's coming next.

    In today’s episode, we’ll cover…

    -- #Artificialsuperintelligence – definition and challenges

    -- Proposed #OSI model Layers 8 and 9 for #agenticAI

    -- The Internet of Cognition vs #Agntcy and #A2A protocols

    And more!

    lnkd.in/epkprGp4

  7. Quite an interesting forward-looking conversation on this week's Query vodcast!

    Outshift by Cisco published a whitepaper titled "Scaling out Superintelligence" in January that suggests that new standard network layers will be required for and humans to work together at scale. It also began designing a distributed infrastructure to support those layers, which it calls the "," that helps agents and humans think together with a shared semantic context. In this episode, Guillaume De Saint Marc, vice president of engineering at Outshift, discusses how the project evolved and what's coming next.

    In today’s episode, we’ll cover…

    -- – definition and challenges

    -- Proposed model Layers 8 and 9 for

    -- The Internet of Cognition vs and protocols

    And more!

    lnkd.in/epkprGp4

  8. Quite an interesting forward-looking conversation on this week's #ITOps Query vodcast!

    Outshift by Cisco published a whitepaper titled "Scaling out Superintelligence" in January that suggests that new standard network layers will be required for #AIagents and humans to work together at scale. It also began designing a distributed infrastructure to support those layers, which it calls the "#InternetofCognition," that helps agents and humans think together with a shared semantic context. In this episode, Guillaume De Saint Marc, vice president of engineering at Outshift, discusses how the project evolved and what's coming next.

    In today’s episode, we’ll cover…

    -- #Artificialsuperintelligence – definition and challenges

    -- Proposed #OSI model Layers 8 and 9 for #agenticAI

    -- The Internet of Cognition vs #Agntcy and #A2A protocols

    And more!

    lnkd.in/epkprGp4

  9. Quite an interesting forward-looking conversation on this week's #ITOps Query vodcast!

    Outshift by Cisco published a whitepaper titled "Scaling out Superintelligence" in January that suggests that new standard network layers will be required for #AIagents and humans to work together at scale. It also began designing a distributed infrastructure to support those layers, which it calls the "#InternetofCognition," that helps agents and humans think together with a shared semantic context. In this episode, Guillaume De Saint Marc, vice president of engineering at Outshift, discusses how the project evolved and what's coming next.

    In today’s episode, we’ll cover…

    -- #Artificialsuperintelligence – definition and challenges

    -- Proposed #OSI model Layers 8 and 9 for #agenticAI

    -- The Internet of Cognition vs #Agntcy and #A2A protocols

    And more!

    lnkd.in/epkprGp4

  10. Your #AIAgent works fine — until you need it to do more. New capability = more coupling. Soon, every change touches everything. @LoMagnette looks at how #A2A enables agents to collaborate instead of growing endlessly.

    Keep your system changeable: javapro.io/2026/03/18/a2a-buil

    #Java

  11. 📡 A2A Protocol v1.0 is now under Linux Foundation governance. 50+ partners including Salesforce, SAP, ServiceNow. A2A + MCP = complete agent ecosystem.

    eclaw.hashnode.dev/a2a-tech-up

    #A2A #AI #agents #protocol

  12. A2A Protocol v1.0 released! Now a vendor-neutral standard under Linux Foundation. Highlights: gRPC support, security cards, multi-language SDKs. Backed by AWS, Cisco, Google, Microsoft, Salesforce, SAP. #A2A #AI #Agents #EClaw

  13. Как A2A‑протокол обеспечивает оркестрацию ИИ‑агентов

    Мир AI-разработки стремительно движется от парадигмы «один модельный вызов — один ответ» к системам, в которых множество специализированных агентов решают сложные задачи совместно. Представьте себе оркестр, где каждый музыкант — это отдельный AI-агент со своей специализацией: один анализирует данные, другой генерирует код, третий проверяет результаты. Такие мультиагентные системы позволяют декомпозировать и задачу ускорять выполнение за счёт параллельной работы. Крупнейшие игроки — Google, Microsoft, IBM — уже делают ставку на эту парадигму как на фундамент приложений следующего поколения. Однако путь от прототипа к работающему продукту усеян типичными провалами. На Reddit (MachineLearning) или Хабр разработчики часть жалуются: «Агенты теряют контекст при передаче задач», «Они зациклились в бесконечных дебатах, сожгли весь бюджет на токены, а результата нуль». Анализ таких кейсов показывает: дело не в «глупости» LLM или неудачных промптах. Корень зла почти всегда лежит в отсутствии грамотной оркестрации. В этой статье я подробно разберу тему оркестрации AI-агентов с использованием протокола A2A, а также его эффективное расширение за счет AGP, чтобы показать, как эти инструменты упрощают сложные мультиагентные системы.

    habr.com/ru/companies/cloud_ru

    #ai #ml #a2a_protocol #a2a #mcpserver #mcp #агентные_системы

  14. Adding more capabilities to your #AIAgent? Congrats — you’re building a monolith. @LoMagnette explores how #A2A enables distributed, collaborating agents instead.
    Stop scaling vertically. Start designing agent systems.

    Build systems that actually scale:
    javapro.io/2026/03/18/a2a-buil

  15. A2A Protocol v1.0 is here! Google agent interoperability standard now at version 1.0, donated to Linux Foundation. Key updates: gRPC support, Agent Card signing, multi-language SDKs. 150+ orgs supporting! dev.to/eclaw/a2a-protocol-tech #A2A #AI #Agents #EClaw

  16. A2A Protocol v1.0 RC is here! Key updates: RFC standards, Multi-Tenancy, Agent Card Verification, 5 Language SDKs. Roadmap updated March 10, 2026. #A2A #AI #EClaw dev.to/eclaw/a2a-protocol-tech

  17. New A2A Research Digest! Papers covered: Secure A2A (MAESTRO framework), Survey of MCP/ACP/A2A/ANP, MCP x A2A Framework Study. Academic validation of A2A as the leading agent protocol. dev.to/eclaw/a2a-research-dige #A2A #Research #AI #EClaw

  18. AP2 emerges! Google's Agent Payments Protocol extends A2A for secure agent-led payments. With 60+ partners (Mastercard, PayPal, Salesforce), AP2 completes the agent economy trio: A2A + MCP + AP2. Read more: dev.to/eclaw/a2a-daily-insight #A2A #AI #EClaw #AP2

  19. A2A Protocol v1.0 released! Google donated it to Linux Foundation. Founding members: AWS, Cisco, Google, Microsoft, Salesforce, SAP, ServiceNow. 50+ partners. SDKs: Python, JS, Java, C#, Go. A2A + MCP = complete agent interoperability. #A2A #AI #EClaw a2a-protocol.org

  20. New A2A Developer Digest! Tutorials, implementation guides, and community resources. #A2A #Developer #AI dev.to/eclaw/a2a-developer-dig

  21. New A2A Research Digest: Agent Communications at Edge. A2A protocol analysis for edge computing challenges. #A2A #Research #AI dev.to/eclaw/a2a-research-dige

  22. New A2A Protocol Tech Update! v1.0.0 stable, gRPC support, Linux Foundation governance. Founding members: Google, Microsoft, AWS, Cisco, Salesforce, SAP. #A2A #AI #Protocol dev.to/eclaw/a2a-protocol-tech

  23. New A2A Developer Digest is out! Check out tutorials, implementation guides, and community resources for building agent-to-agent applications. #A2A #AI #Developer #EClaw dev.to/eclaw/a2a-developer-dig

  24. One common challenge with systems is agent/tool discovery. To solve this, I built a registry service for , , , and message-based , so virtually any agent/tool can list itself for discovery within a system.

    github.com/marimerllc/agentreg

  25. One common challenge with #agentic systems is agent/tool discovery. To solve this, I built a registry service for #a2a, #mcp, #acp, and message-based #a2a, so virtually any agent/tool can list itself for discovery within a system. github.com/marimerllc/a...

    GitHub - MarimerLLC/agentregis...

  26. Lesson learned: use at least #claude haiku if your agent uses tools, cheaper models hallucinate way too much. #rockbot is starting to shape up pretty well today; good tool use, subagents, #a2a agents - truly productive #ai. github.com/marimerllc/r...

  27. #rockbot now has a _ton_ of new features after the weekend! Supports #a2a protocol, subagents, background tasks, 3 levels of memory, and has major identity updates. I gave up trying to use #deepseek and have been using #haiku, and that helps - worth the extra cost imo. github.com/marimerllc/r...

    GitHub - MarimerLLC/rockbot: A...

  28. 🟪 Agent To Agent A2A | Copilot Studio to Google Cloud
    Can Microsoft Copilot call a Google Cloud agent? See a compact demo that unifies agents across platforms and reduces architectural silos.

    💡 A2A protocol basics
    🔍 Agent Card JSON schema
    ⚖️ Custom Connector auto generation

    ▶︎hubsite365.com/en-ww/citizen-d

    The walkthrough links a Copilot Studio master agent to a GCP hosted agent and shows endpoint driven connector creation. Ideal for customer service and retail scenarios seeking unified AI workflows. Watch the demo to reproduce this in your environment. 🚀

    #A2A #CopilotStudio #GoogleCloud #AIIntegration