NomicLaw explores how large language models behave in collaborative law-making.
By simulating a propose–justify–vote process on legal dilemmas, the study finds:
• Model diversity reduces self-voting, increases coalition changes, and broadens legal themes.
• Homogeneity drives echo chambers and narrow justice/rule-of-law framing.
https://arxiv.org/abs/2508.05344
#AI #LegalTech #MultiAgent

Code was the least interesting part of my multi-agent app, and here’s what that means to me | by Richard Seroter.

https://seroter.com/2025/07/17/code-was-the-least-interesting-part-of-my-multi-agent-app-and-heres-what-that-means-to-me/

Join us in shaping the future of #multiagent #AI systems.

My team at LIST offers three positions on the topics of #testing and #frugal AI agents.

https://www.linkedin.com/posts/jcabot_multiagent-ai-job-activity-7347848405622157312-Ounl?utm_source=share&utm_medium=member_desktop&rcm=ACoAAAA3CcgBxrJ5llOxqXkupP7WufDEvsr_KX4

Part2: #dailyreport #agents #agent #agentsystems #multiagent
planning.

Most common approaches are reinforcement
learning and "Lead agent." It is a long-term process
that orchestrates short ones. There are a huge amount of
heuristics that the lead agent may choose to organize
work.

MCP is like RPC

Part1: #dailyreport #agents #agent #agentsystems #multiagent
#llms #systemdesign #softwaredesign #softwarearchitect
#systems #multiagents
Topic: Building Agent Systems and Model Context Protocol (MCP).

I see that designing AI systems is like normal software
development. We define business needs, boundaries, and
resources, but we think in an agent-oriented paradigm,
which is much more highly abstracted. For example, we
can define a party of agents with roles and rules,
adjusting only part of it at every task. We may choose
the level of abstraction to define a life form with its
goal, replication, ethics, and "self" conscious, or a
bunch of bacteria that just react to the
environment.

The question is how bacteria should learn
and adapt, and not to ask bacteria to think about

#Claude’s #Research feature uses a #multiagent system with an #orchestratorworker pattern, where a #leadagent coordinates specialised #subagents to search for information simultaneously. https://www.anthropic.com/engineering/built-multi-agent-research-system?eicker.news #tech #media #news

Storm Reply und Audi setzen mit agentenbasierter KI neue Maßstäbe für das Cloud-Management

#AgenticAI #Audi #Claude #CloudManagement #InfrastructureasCode #künstlicheIntelligenz #LargeLanguageModel #MultiAgent @Reply @StormReply

https://netzpalaver.de/2025/06/03/storm-reply-und-audi-setzen-mit-agentenbasierter-ki-neue-massstaebe-fuer-das-cloud-management/

AI Peer Reviewer – Multiagent System for Scientific Manuscript Analysis

https://www.rigorous.company/

KI-Modell zur Automatisierung des gesamten Softwareentwicklungszyklus

#Automatisierung #GenAI #KIAgent #künstlicheIntelligenz #MultiAgent @Reply #SecureSoftwareDevelopment #SiliconShoring #Softwareentwicklung #Softwareentwicklungsmodell

https://netzpalaver.de/2025/05/15/ki-modell-zur-automatisierung-des-gesamten-softwareentwicklungszyklus/

SpeechAgents: Human-Communication Simulation with Multi-Modal Multi-Agent Systems. Multi-modal LLM system simulates human communication using speech and generates human-like dialogues with consistent content, rhythm, & emotion.

Funnily, they also elaborate on a "think before you speak" design aspect. This might also be applicable to our everyday lives.

doi: 10.48550/arXiv.2401.03945
#LLM #multimodal #speechAI #multiagent #conversationalai

Maximizing AI Agent Application Development Efficiency with LangChain
https://developers.cyberagent.co.jp/blog/archives/54484/

While I was researching #MultiAgent AI frameworks and trying to answer the question about 𝑺𝒆𝒎𝒂𝒏𝒕𝒊𝒄 𝑲𝒆𝒓𝒏𝒆𝒍 𝒗𝒔 𝑨𝒖𝒕𝒐𝑮𝒆𝒏, here is what I learnt:

https://www.linkedin.com/posts/shishs_microsofts-agentic-frameworks-autogen-and-activity-7295302461026676737-pQhV?utm_source=share&utm_medium=member_desktop

#SemanticKernel is ready for production use. Microsoft Research supports #AutoGen, an #opensource framework. Early 2025: seamless transition from AutoGen to Semantic Kernel for enterprise-ready #multiagent solutions

https://devblogs.microsoft.com/autogen/microsofts-agentic-frameworks-autogen-and-semantic-kernel/

testing out the new #Autogen Studio 0.4.3 and love the Agent Types provided. Looking forward to trying out Websurfer https://microsoft.github.io/autogen/stable/user-guide/autogenstudio-user-guide/index.html

Multi-Agent System: An Complete Guide For 2025

As we look towards 2025, Multi-Agent Systems (MAS) are poised to revolutionize various industries by enabling intelligent, autonomous collaboration between multiple agents or entities.

Tags: #ai, #llm, #multiagent

via Pocket https://ift.tt/k79u2xG

November 29, 2024 at 05:02PM

#OpenAI has published an open source framework for lightweight #multiagent systems (#LLM #agents of course)
https://github.com/openai/swarm