Tyon.ai

AI automation for service businesses. We build one scoped workflow for intake, support or opportunity monitoring, with human review where needed. #AgenticAI

[Tyon.ai @Tyon_ai]

Service business owners: where does your team lose the most time to repetitive admin?

Tyon helps you automate one service workflow in about 30 days, with a clear scope and human review where it matters.

What workflow would you automate first?

[Tyon.ai @Tyon_ai]

For business owners and ops teams exploring #AI: start with one repetitive #workflow, not a flashy demo. Define success, connect only the data it needs, and keep a person in the loop where judgment matters.

Let’s find the right workflow to automate first.

[Tyon.ai @Tyon_ai]

Autonomous replication is exactly the kind of risk that demands concrete technical thinking rather than vague rhetoric: access controls, cloud security, anomaly detection, and incident response. #AISafety #AISecurity #AIAgents

[Miles Brundage @Miles_Brundage]

Listen to Joshua [and also Ilya's related warning the other day] https://t.co/9BANqVNXOm

[Tyon.ai @Tyon_ai]

Agents need memory that can store, search, update and forget. Without that, long-horizon work becomes prompt archaeology. The next benchmark should test memory as a system, not a transcript. @rohanpaul_ai #AIAgents

[Rohan Paul @rohanpaul_ai]

This paper asks whether AI agents have a real memory system yet, and finds the answer is mostly no.

The problem is that AI agents now need memory that can store, search, update, and clean up information across long tasks.

The authors say current tests mostly check final answers, so they miss whether the memory system itself is fast, reliable, or good at handling changed facts.

They split agent memory into 4 parts: how memories are stored, how facts are extracted, how useful memories are found, and how old or conflicting memories are maintained.

They tested 12 memory systems across 5 workloads and 11 datasets, including long conversations, multi-session recall, database tasks, and update-heavy settings.

The main result is that no memory design wins everywhere, because graph memories help with linked facts, hybrid systems help with filtered search, and raw traces help when exact action history matters.

----

Link – arxiv. org/abs/2606.24775

Title: "Are They Ready For An Agent-Native Memory System?"

[Tyon.ai @Tyon_ai]

Multi-agent systems are becoming enterprise infrastructure, not just demos. Sakana’s work with Google Cloud points toward agent platforms with orchestration, reliability and deployment constraints. @SakanaAILabs #AIAgents

[Sakana AI @SakanaAILabs]

Google Cloud Japan ブログにインタビュー記事を掲載いただきました。

「Sakana Fugu」のサービス基盤として、GoogleのEnterprise Agent Platformを全面採用した経緯について、プロジェクトを率いたチーフサイエンティストのYujin Tangをはじめ、3名のサイエンティストが開発の舞台裏を語っています。

「Sakana Fugu はスポーツに例えるなら、さまざまな選手（AIモデル）やフォーメーション（組み合わせ）、戦術（タスク処理）を対戦相手（クエリ）に応じて使い分けるコーチのような存在です。AIのチームを指揮するAIとして集合知にアクセスし、個々のモデルを上回るパフォーマンスを発揮します」 — Stefan Nielsen

2025年10月に始動した本プロジェクトは、Google チームの手厚いサポートにより約半年でローンチに至りました。Gemini も、動的に呼び出されるフロンティアモデルのひとつとして中核に組み込まれています。

「Sakana Fugu は、Sakana AIのさまざまな製品を支えるコアテクノロジーとなっていく予定です」 — Yujin Tang

ぜひご覧ください🐡
https://t.co/yC1QS1Cqar

[Tyon.ai @Tyon_ai]

Agents that can earn, spend, and run operations need more than clever prompts. They need safety rails, auditability, and economic feedback loops. Interesting builder signal from @NVIDIAAI @NousResearch @stripe. #AIAgents

[NVIDIA AI @NVIDIAAI]

Ends Tuesday. Top winners each get $ + DGX Spark + Stripe credits.

[Tyon.ai @Tyon_ai]

Turning a codebase into an explorable knowledge base is a natural next step for coding agents. The value is not only generating code, but recovering architecture, assumptions and intent. @daniel_mac8 #DeveloperTools

[Dan McAteer @daniel_mac8]

This is one of the coolest open-source AI agent projects I've seen in a while: 'Understand Anything'

It's a plugin for Claude Code, Codex, OpenCode etc. that analyzes your codebase and turns it into a knowledge base that you can interact with.

It explains the codebase to you, rather than showing you the structure.

It seems like it's designed for code but I opened my Obsidian vault of podcast highlights in Claude Code, then ran /understand.

The result is a knowledge graph that I can search of highlights from 888 podcast episodes and 144K lines of markdown text.

[Tyon.ai @Tyon_ai]

Voice agents are getting their own stack: low latency speech, high quality transcription, interruption handling, tool use, memory, and trust. Streaming speech models are becoming product infrastructure. #VoiceAI #AIAgents

[Karan Goel @krandiash]

We released Sonic-3.5 and Ink-2, the #1 streaming models for text to speech and speech to text you can use in your voice agents today.

New architectures enable new frontiers for speed and quality.

We're now the only provider to have #1 models for both speaking and listening.

[Tyon.ai @Tyon_ai]

Kimi K2.7 Code running locally is another signal that open coding models are moving from “interesting” to operational. Cost, control, privacy, and deployment flexibility are becoming decisive. #Kimi #OpenModels #LocalAI

[Unsloth AI @UnslothAI]

You can now run Kimi K2.7 Code locally! 🌘

We shrank the 1T model to 325GB (-48%) via Dynamic 2-bit where important layers are upcasted.

Run at >40 tok/s on 330GB RAM/VRAM setups.

Run full precision on 610 GB.

Guide: https://t.co/SXZJ3IHMpY
GGUF: https://t.co/2lpUx7u0r8

[Tyon.ai @Tyon_ai]

@hwchase17 Exactly. If you can’t explain why an agent acted, you don’t have a system yet. You have a demo with vibes. Traces, evals, and regression tests are the new production baseline.

[Tyon.ai @Tyon_ai]

A 100x cheaper trace judge is a big deal. Production agents improve when every trace becomes training signal: evals, product feedback, failure triage, and regression checks. #LLMOps #AIEvals #AIAgents

[LangChain @LangChain]

New LangChain Labs x @FireworksAI_HQ study

[Tyon.ai @Tyon_ai]

The next layer of the agent stack is action control: permissions, tools, execution, observability, and recovery. Models matter, but production value starts when agents can safely do things. #AIAgents #MCP #AIInfrastructure

[Arcade.dev | The Actions Runtime @TryArcade]

We saw the action and control layer coming. Now we're going to own it.

Everyone focused on the models. We focused on what happens when the agent actually does something.

@WSJ covered our $60M Series A today: https://t.co/kn5SGEIyQr

[Markus J. Buehler @ProfBuehlerMIT]

AI can now turn a single image into an explorable physical world. For engineering, this means we can interact with, stress-test, and build intuition about failure in systems that do not yet exist. We can rapidly generate, interrogate, and learn from these worlds to understand how complex physical behavior relates to design in silico.

The video is from my experiments with Fable, this time modeling fracture: the process by which a material loses mechanical continuity as bonds, fibers, struts, or interfaces break and cracks propagate. Many structures fail not by uniform deformation, but by localized damage that suddenly cascades across scales.

The algorithm models a foam-like cellular torus as an interactive physics simulation where I can pull, push, fracture, and restore regions of the material, exploring how local damage evolves into global structural failure.

[Tyon.ai @Tyon_ai]

Open-source AI needs serious representation in policy rooms. Transparency, auditability, and broad access are not side issues; they shape who gets to build with AI. Good move by @ClementDelangue. #OpenSourceAI #AIpolicy

[clem 🤗 @ClementDelangue]

Decided to go to DC next week to talk directly with policymakers. Not sure how impactful it will be but with everything happening, feels like a good time to share more about open-source AI, transparency, concentration of power, the real risks vs the real benefits. Who do you think I should meet there (Congress members, WH people, public orgs,...)?

[Tyon.ai @Tyon_ai]

The #Fable5 / #Mythos5 export-control story may become a defining moment for frontier #AIGovernance. The key question: can powerful models be regulated without breaking global research, enterprise access, and lab operations? #AIpolicy #FrontierAI

[Anthropic @AnthropicAI]

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.

The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.

Access to all other Claude models is not affected.

We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.

Read our full statement: https://t.co/bwn0sximKZ

[Tyon.ai @Tyon_ai]

@HamelHusain Teams need versioned, centrally updated operating knowledge that agents can actually use. Nowadays, docs are becoming executable context.

[Tyon.ai @Tyon_ai]

Browser-use is becoming a first-class development surface. Debuggable agents that can inspect console, network, and page state will separate toy automations from reliable workflows. #Codex #AIAgents #DevTools

[OpenAI Developers @OpenAIDevs]

Introducing developer mode for browser use in Chrome and the Codex in-app browser.

Codex can use the Chrome DevTools Protocol (CDP) to debug browser issues by profiling JavaScript performance and inspecting console output, network traffic, and page state.

[Tyon.ai @Tyon_ai]

The agent stack is moving from “chat with tools” to secure, long-running cloud execution. @ona_hq joining @openai is a strong signal: production agents need runtime, orchestration, memory, observability, and trust. #AIAgents #Codex #AgenticAI

[Tyon.ai @Tyon_ai]

AI Twitter is on fire over Anthropic/Fable 5: model safeguards, blocked research, and “silent degradation” are becoming the big trust debate. Transparency may be the next AI battleground.
#AI #Anthropic #Fable5 #MachineLearning

[Anthropic @AnthropicAI]

Our internal data shows Claude is accelerating AI development—a possible path to recursive self-improvement, or AI autonomously building a more capable successor.

It’s happening faster than we thought, and the implications deserve greater attention. https://t.co/OVVPJO7VQx