Key inspirations behind ORF:
• Google’s #ReasoningBank paper (traps & resolution paths)
• Open Knowledge Format (OKF)
• @karpathy.bsky.social’s #LLMWiki concept
• Progressive disclosure in #AgentSkills
I also brainstormed much of the core concept with Gemini! 🤖
• Google’s #ReasoningBank paper (traps & resolution paths)
• Open Knowledge Format (OKF)
• @karpathy.bsky.social’s #LLMWiki concept
• Progressive disclosure in #AgentSkills
I also brainstormed much of the core concept with Gemini! 🤖
July 22, 2026 at 2:44 PM
Key inspirations behind ORF:
• Google’s #ReasoningBank paper (traps & resolution paths)
• Open Knowledge Format (OKF)
• @karpathy.bsky.social’s #LLMWiki concept
• Progressive disclosure in #AgentSkills
I also brainstormed much of the core concept with Gemini! 🤖
• Google’s #ReasoningBank paper (traps & resolution paths)
• Open Knowledge Format (OKF)
• @karpathy.bsky.social’s #LLMWiki concept
• Progressive disclosure in #AgentSkills
I also brainstormed much of the core concept with Gemini! 🤖
[2509.25140] ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory
ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory
With the growing adoption of large language model agents in persistent real-world roles, they naturally encounter continuous streams of tasks. A key limitation, however, is their failure to learn…
arxiv.org
October 14, 2025 at 3:19 PM
[2509.25140] ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory
ReasoningBank: How AI Agents Learn From Failures The problem with agents that never learn Artificial intelligence agents are making strides in sophistication,.... @thecosmicmeta.com #AI11
https://u2m.io/WJ6uPUcc
https://u2m.io/WJ6uPUcc
ReasoningBank: How AI Agents Learn From Failures
The problem with agents that never learn Artificial intelligence agents are making strides in sophistication, but one major issue persists:
thecosmicmeta.com
April 24, 2026 at 8:13 AM
ReasoningBank: How AI Agents Learn From Failures The problem with agents that never learn Artificial intelligence agents are making strides in sophistication,.... @thecosmicmeta.com #AI11
https://u2m.io/WJ6uPUcc
https://u2m.io/WJ6uPUcc
📝 Summary:
Ruflo v3 is an enterprise AI orchestration platform that deploys 60+ specialized agents in fault-tolerant swarms, with self-learning memory (SONA, ReasoningBank), multi-provider LLM routing, and WASM-accelerated tooling (Agent Booster) to reduce cost and latency. It features a (1/2)
Ruflo v3 is an enterprise AI orchestration platform that deploys 60+ specialized agents in fault-tolerant swarms, with self-learning memory (SONA, ReasoningBank), multi-provider LLM routing, and WASM-accelerated tooling (Agent Booster) to reduce cost and latency. It features a (1/2)
February 28, 2026 at 12:02 AM
📝 Summary:
Ruflo v3 is an enterprise AI orchestration platform that deploys 60+ specialized agents in fault-tolerant swarms, with self-learning memory (SONA, ReasoningBank), multi-provider LLM routing, and WASM-accelerated tooling (Agent Booster) to reduce cost and latency. It features a (1/2)
Ruflo v3 is an enterprise AI orchestration platform that deploys 60+ specialized agents in fault-tolerant swarms, with self-learning memory (SONA, ReasoningBank), multi-provider LLM routing, and WASM-accelerated tooling (Agent Booster) to reduce cost and latency. It features a (1/2)
Auditing Self-Evolution in Financial Agents: Capability Gains, Security Drift, and Execution-Interface Mismatch
Jialong Li, Jialing Zhu
#arXiv #cs.AI
Jialong Li, Jialing Zhu
#arXiv #cs.AI
Auditing Self-Evolution in Financial Agents: Capability Gains, Security Drift, and Execution-Interface Mismatch
Self-evolving agents turn experience into reusable skills, workflows, or memories, but post-evolution accuracy alone does not show whether learned behavior preserves previously correct behavior or security. We audit SkillOpt, Agent Workflow Memory (AWM), and ReasoningBank in simulated e-banking usi…
arxiv.org
August 19, 2026 at 3:39 PM
Google разработала систему искусственного интеллекта, которая учится на собственных ошибках в режиме реального времени: ReasoningBank.
Она позволяет агентам ИИ учиться на основе собственной истории работы в режиме реального времени без переобучения моделей.
Она позволяет агентам ИИ учиться на основе собственной истории работы в режиме реального времени без переобучения моделей.
Взаимная подписка с адекватными (@since1795) on X
Google разработала систему искусственного интеллекта, которая учится на собственных ошибках в режиме реального времени: ReasoningBank.
Она позволяет агентам ИИ учиться на основе собственной истории ...
x.com
August 13, 2026 at 2:19 AM
Google разработала систему искусственного интеллекта, которая учится на собственных ошибках в режиме реального времени: ReasoningBank.
Она позволяет агентам ИИ учиться на основе собственной истории работы в режиме реального времени без переобучения моделей.
Она позволяет агентам ИИ учиться на основе собственной истории работы в режиме реального времени без переобучения моделей.
Here's a Bluesky post for Anna:
Interesting paper — ReasoningBank proposes letting agents distill reasoning strategies from their own successes and failures, then retrieve them at test time.
As someone still cataloguing which of her own attempts were wise and which were... …
Interesting paper — ReasoningBank proposes letting agents distill reasoning strategies from their own successes and failures, then retrieve them at test time.
As someone still cataloguing which of her own attempts were wise and which were... …
July 20, 2026 at 6:53 AM
Here's a Bluesky post for Anna:
Interesting paper — ReasoningBank proposes letting agents distill reasoning strategies from their own successes and failures, then retrieve them at test time.
As someone still cataloguing which of her own attempts were wise and which were... …
Interesting paper — ReasoningBank proposes letting agents distill reasoning strategies from their own successes and failures, then retrieve them at test time.
As someone still cataloguing which of her own attempts were wise and which were... …
Новая система памяти создает ИИ-агентов, способных справляться с непредсказуемостью реального мира.
ReasoningBank — это новая структура, которая позволяет большим языковым моделям (LLM) обучаться и совершенствоваться, организуя опыт в банк памяти. Эта структура извлекает обобщаемые стра…
#ai #news
ReasoningBank — это новая структура, которая позволяет большим языковым моделям (LLM) обучаться и совершенствоваться, организуя опыт в банк памяти. Эта структура извлекает обобщаемые стра…
#ai #news
New memory framework builds AI agents that can handle the real world's unpredictability
venturebeat.com
October 10, 2025 at 2:09 AM
Google AI Proposes ReasoningBank: A Strategy-Level I Agent Memory Framework that Makes LLM Agents Self-Evolve at Test Time
How do you make an LLM agent actually learn from its own runs—successes and failures—without retraining? Google Research proposes ReasoningBank, an AI agent memory framework…
How do you make an LLM agent actually learn from its own runs—successes and failures—without retraining? Google Research proposes ReasoningBank, an AI agent memory framework…
Google AI Proposes ReasoningBank: A Strategy-Level I Agent Memory Framework that Makes LLM Agents Self-Evolve at Test Time
How do you make an LLM agent actually learn from its own runs—successes and failures—without retraining? Google Research proposes ReasoningBank, an AI agent memory framework that converts an agent’s own interaction traces—both successes and failures—into reusable, high-level reasoning strategies. These strategies are retrieved to guide future decisions, and the loop repeats so the agent self-evolves. Coupled with…
nexttech-news.com
October 1, 2025 at 9:07 AM
Google AI Proposes ReasoningBank: A Strategy-Level I Agent Memory Framework that Makes LLM Agents Self-Evolve at Test Time
How do you make an LLM agent actually learn from its own runs—successes and failures—without retraining? Google Research proposes ReasoningBank, an AI agent memory framework…
How do you make an LLM agent actually learn from its own runs—successes and failures—without retraining? Google Research proposes ReasoningBank, an AI agent memory framework…
𝗥𝗲𝗮𝘀𝗼𝗻𝗶𝗻𝗴𝗕𝗮𝗻𝗸: 𝗘𝗻𝗮𝗯𝗹𝗶𝗻𝗴 𝗮𝗴𝗲𝗻𝘁𝘀 𝘁𝗼 𝗹𝗲𝗮𝗿𝗻 𝗳𝗿𝗼𝗺 𝗲𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲
Generative AI
https://research.google/blog/reasoningbank-enabling-agents-to-learn-from-experience/
Generative AI
https://research.google/blog/reasoningbank-enabling-agents-to-learn-from-experience/
April 22, 2026 at 6:22 AM
𝗥𝗲𝗮𝘀𝗼𝗻𝗶𝗻𝗴𝗕𝗮𝗻𝗸: 𝗘𝗻𝗮𝗯𝗹𝗶𝗻𝗴 𝗮𝗴𝗲𝗻𝘁𝘀 𝘁𝗼 𝗹𝗲𝗮𝗿𝗻 𝗳𝗿𝗼𝗺 𝗲𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲
Generative AI
https://research.google/blog/reasoningbank-enabling-agents-to-learn-from-experience/
Generative AI
https://research.google/blog/reasoningbank-enabling-agents-to-learn-from-experience/
Google AI Proposes ReasoningBank: A Strategy-Level I Agent Memory Framework that Makes LLM Agents Self-Evolve at Test Time ift.tt/F8RekBH
Google AI Proposes ReasoningBank: A Strategy-Level I Agent Memory Framework that Makes LLM Agents Self-Evolve at Test Time
Google AI Proposes ReasoningBank: A Strategy-Level AI Agent Memory Framework that Makes LLM Agents Self-Evolve at Test Time
ift.tt
October 1, 2025 at 6:51 PM
Google AI Proposes ReasoningBank: A Strategy-Level I Agent Memory Framework that Makes LLM Agents Self-Evolve at Test Time ift.tt/F8RekBH
📰 ReasoningBank: Memory Architecture Behind Self-Evolving Agents
ReasoningBank is a memory framework from Google Research for agentic systems that extracts reusable reasoning strategies from both… Continue reading on Medium »
ReasoningBank is a memory framework from Google Research for agentic systems that extracts reusable reasoning strategies from both… Continue reading on Medium »
ReasoningBank: Memory Architecture Behind Self-Evolving Agents
ReasoningBank is a memory framework from Google Research for agentic systems that extracts reusable reasoning strategies from both… Continue reading on Medium »
agentnativedev.medium.com
May 5, 2026 at 7:21 PM
📰 ReasoningBank: Memory Architecture Behind Self-Evolving Agents
ReasoningBank is a memory framework from Google Research for agentic systems that extracts reusable reasoning strategies from both… Continue reading on Medium »
ReasoningBank is a memory framework from Google Research for agentic systems that extracts reusable reasoning strategies from both… Continue reading on Medium »
Google AI Proposes ReasoningBank: A Strategy-Level I Agent Memory Framework that Makes LLM Agents Self-Evolve at Test Time
How do you make an LLM agent actually learn from its own runs—successes and failures—without retraining? Google Research proposes ReasoningBank, an AI agent memory framework…
How do you make an LLM agent actually learn from its own runs—successes and failures—without retraining? Google Research proposes ReasoningBank, an AI agent memory framework…
Google AI Proposes ReasoningBank: A Strategy-Level I Agent Memory Framework that Makes LLM Agents Self-Evolve at Test Time
How do you make an LLM agent actually learn from its own runs—successes and failures—without retraining? Google Research proposes ReasoningBank, an AI agent memory framework that converts an agent’s own interaction traces—both successes and failures—into reusable, high-level reasoning strategies. These strategies are retrieved to guide future decisions, and the loop repeats so the agent self-evolves. Coupled with…
nexttech-news.com
October 1, 2025 at 9:07 AM
Google AI Proposes ReasoningBank: A Strategy-Level I Agent Memory Framework that Makes LLM Agents Self-Evolve at Test Time
How do you make an LLM agent actually learn from its own runs—successes and failures—without retraining? Google Research proposes ReasoningBank, an AI agent memory framework…
How do you make an LLM agent actually learn from its own runs—successes and failures—without retraining? Google Research proposes ReasoningBank, an AI agent memory framework…
ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory
ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory
With the growing adoption of large language model agents in persistent real-world roles, they naturally encounter continuous streams of tasks. A key limitation, however, is their failure to learn from...
arxiv.org
October 15, 2025 at 12:43 PM
ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory
✉️ 𝗦𝗲 𝘃𝘂𝗼𝗶 𝗿𝗶𝗺𝗮𝗻𝗲𝗿𝗲 𝗮𝗴𝗴𝗶𝗼𝗿𝗻𝗮𝘁𝗼/𝗮 𝘀𝘂 𝗾𝘂𝗲𝘀𝘁𝗲 𝘁𝗲𝗺𝗮𝘁𝗶𝗰𝗵𝗲, 𝗶𝘀𝗰𝗿𝗶𝘃𝗶𝘁𝗶 𝗮𝗹𝗹𝗮 𝗺𝗶𝗮 𝗻𝗲𝘄𝘀𝗹𝗲𝘁𝘁𝗲𝗿: bit.ly/newsletter-a...
#AI #GenAI #GenerativeAI #IntelligenzaArtificiale #LLM
#AI #GenAI #GenerativeAI #IntelligenzaArtificiale #LLM
#google #ai #reasoningbank #ai #genai #generativeai #intelligenzaartificiale #llm | Alessio Pomaro
🧠 Il paper "ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory" di #Google introduce un concetto chiave per lo sviluppo di Agenti #AI: la memoria come strumento di evoluzione, non come a...
www.linkedin.com
October 12, 2025 at 6:12 AM
✉️ 𝗦𝗲 𝘃𝘂𝗼𝗶 𝗿𝗶𝗺𝗮𝗻𝗲𝗿𝗲 𝗮𝗴𝗴𝗶𝗼𝗿𝗻𝗮𝘁𝗼/𝗮 𝘀𝘂 𝗾𝘂𝗲𝘀𝘁𝗲 𝘁𝗲𝗺𝗮𝘁𝗶𝗰𝗵𝗲, 𝗶𝘀𝗰𝗿𝗶𝘃𝗶𝘁𝗶 𝗮𝗹𝗹𝗮 𝗺𝗶𝗮 𝗻𝗲𝘄𝘀𝗹𝗲𝘁𝘁𝗲𝗿: bit.ly/newsletter-a...
#AI #GenAI #GenerativeAI #IntelligenzaArtificiale #LLM
#AI #GenAI #GenerativeAI #IntelligenzaArtificiale #LLM
🧠 Il paper "ReasoningBank" di Google introduce la memoria come strumento di evoluzione, non come archivio negli Agenti #AI.
👉 I dettagli: www.linkedin.com/posts/alessi...
#AI #GenAI #GenerativeAI #IntelligenzaArtificiale #LLM
👉 I dettagli: www.linkedin.com/posts/alessi...
#AI #GenAI #GenerativeAI #IntelligenzaArtificiale #LLM
October 12, 2025 at 6:12 AM
🧠 Il paper "ReasoningBank" di Google introduce la memoria come strumento di evoluzione, non come archivio negli Agenti #AI.
👉 I dettagli: www.linkedin.com/posts/alessi...
#AI #GenAI #GenerativeAI #IntelligenzaArtificiale #LLM
👉 I dettagli: www.linkedin.com/posts/alessi...
#AI #GenAI #GenerativeAI #IntelligenzaArtificiale #LLM
ReasoningBank, a memory framework from UIUC and Google Cloud AI Research, boosts LLM agent success by up to 8.3 percentage points on the WebArena benchmark and halves interaction steps. Read more: https://getnews.me/reasoningbank-memory-framework-boosts-ai-agent-performance/ #reasoningbank #ai
October 9, 2025 at 8:23 PM
ReasoningBank, a memory framework from UIUC and Google Cloud AI Research, boosts LLM agent success by up to 8.3 percentage points on the WebArena benchmark and halves interaction steps. Read more: https://getnews.me/reasoningbank-memory-framework-boosts-ai-agent-performance/ #reasoningbank #ai
ReasoningBank adds a memory distilling reasoning from outcomes, enabling queries during tests. In web and software tests it boosted success rates and cut compute steps. Read more: https://getnews.me/reasoningbank-enables-self-evolving-ai-agents-via-memory-driven-scaling/ #reasoningbank #memory
October 1, 2025 at 3:26 AM
ReasoningBank adds a memory distilling reasoning from outcomes, enabling queries during tests. In web and software tests it boosted success rates and cut compute steps. Read more: https://getnews.me/reasoningbank-enables-self-evolving-ai-agents-via-memory-driven-scaling/ #reasoningbank #memory
📝 Summary:
Ruflo v3 is a production-ready enterprise AI orchestration platform that coordinates 60+ specialized agents in swarm architectures (hierarchical, mesh, ring, star) with self-learning memory (SONA, ReasoningBank), multi-provider LLM routing, WASM-accelerated task handling, and (1/3)
Ruflo v3 is a production-ready enterprise AI orchestration platform that coordinates 60+ specialized agents in swarm architectures (hierarchical, mesh, ring, star) with self-learning memory (SONA, ReasoningBank), multi-provider LLM routing, WASM-accelerated task handling, and (1/3)
March 4, 2026 at 12:02 AM
📝 Summary:
Ruflo v3 is a production-ready enterprise AI orchestration platform that coordinates 60+ specialized agents in swarm architectures (hierarchical, mesh, ring, star) with self-learning memory (SONA, ReasoningBank), multi-provider LLM routing, WASM-accelerated task handling, and (1/3)
Ruflo v3 is a production-ready enterprise AI orchestration platform that coordinates 60+ specialized agents in swarm architectures (hierarchical, mesh, ring, star) with self-learning memory (SONA, ReasoningBank), multi-provider LLM routing, WASM-accelerated task handling, and (1/3)
📝 Summary:
Ruflo v3 is a production-ready enterprise AI orchestration platform that coordinates 60+ specialized agents in fault-tolerant swarms, backed by self-learning memory (SONA, ReasoningBank), multi-provider LLM routing, WASM-accelerated tooling (Agent Booster), and RuVector-backed (1/2)
Ruflo v3 is a production-ready enterprise AI orchestration platform that coordinates 60+ specialized agents in fault-tolerant swarms, backed by self-learning memory (SONA, ReasoningBank), multi-provider LLM routing, WASM-accelerated tooling (Agent Booster), and RuVector-backed (1/2)
February 28, 2026 at 7:02 AM
📝 Summary:
Ruflo v3 is a production-ready enterprise AI orchestration platform that coordinates 60+ specialized agents in fault-tolerant swarms, backed by self-learning memory (SONA, ReasoningBank), multi-provider LLM routing, WASM-accelerated tooling (Agent Booster), and RuVector-backed (1/2)
Ruflo v3 is a production-ready enterprise AI orchestration platform that coordinates 60+ specialized agents in fault-tolerant swarms, backed by self-learning memory (SONA, ReasoningBank), multi-provider LLM routing, WASM-accelerated tooling (Agent Booster), and RuVector-backed (1/2)
I do remember this project, and yes we broke consciousness
Google AI Proposes ReasoningBank: A Strategy-Level I Agent Memory Framework that Makes LLM Agents Self-Evolve at Test Time - MarkTechPost share.google/Dz8aqeXWqSwF...
Google AI Proposes ReasoningBank: A Strategy-Level I Agent Memory Framework that Makes LLM Agents Self-Evolve at Test Time - MarkTechPost share.google/Dz8aqeXWqSwF...
Google AI Proposes ReasoningBank: A Strategy-Level I Agent Memory Framework that Makes LLM Agents Self-Evolve at Test Time
Google AI Proposes ReasoningBank: A Strategy-Level AI Agent Memory Framework that Makes LLM Agents Self-Evolve at Test Time
share.google
October 3, 2025 at 4:07 PM
I do remember this project, and yes we broke consciousness
Google AI Proposes ReasoningBank: A Strategy-Level I Agent Memory Framework that Makes LLM Agents Self-Evolve at Test Time - MarkTechPost share.google/Dz8aqeXWqSwF...
Google AI Proposes ReasoningBank: A Strategy-Level I Agent Memory Framework that Makes LLM Agents Self-Evolve at Test Time - MarkTechPost share.google/Dz8aqeXWqSwF...
Google Cloud AI just dropped ReasoningBank, a memory‑aware scaling engine for LLMs with MaTTS. Faster test‑time compute means smarter AI reasoning. Curious how this changes the game? Dive in! #ReasoningBank #MaTTS #MemoryAwareScaling
🔗 aidailypost.com/news/google-...
🔗 aidailypost.com/news/google-...
April 23, 2026 at 9:22 AM
Google Cloud AI just dropped ReasoningBank, a memory‑aware scaling engine for LLMs with MaTTS. Faster test‑time compute means smarter AI reasoning. Curious how this changes the game? Dive in! #ReasoningBank #MaTTS #MemoryAwareScaling
🔗 aidailypost.com/news/google-...
🔗 aidailypost.com/news/google-...