A ticket-booking agent trained in a mock app improved real-phone success from 28.8% to 48.1%.
https://oossa.com/en/mai-gym-platform-released-for-training-and-testing-gui-ai-agents
@oossa.bsky.social
@oossa.bsky.socialA ticket-booking agent trained in a mock app improved real-phone success from 28.8% to 48.1%.
https://oossa.com/en/mai-gym-platform-released-for-training-and-testing-gui-ai-agents
@inautilo.bsky.social#Development #Previews
Modernizing the Visual Studio Code UI · Where coding and agents finally feel alike ilo.im/16h1hx by Lee Murray
_____
#Microsoft #VsCode #CodeEditor #AI #AiAgents #DevOps #WebDev #Frontend #Backend
@puretech.newsPerplexity AI open-sources its secure AI agent sandbox SPACE #Technology #AI #AIAgents #perplexityai #aisandbox #aiagentsecurity #autonomoussystems #opensource
@ossradarai.bsky.socialAlphaDiana is a new arXiv paper introducing a modular measurement system for evaluating AI agents on verifiable reasoning tasks, separating harness control loops from file reuse and memory retention. It benchmarks three…
#OpenSource #AIagents #Benchmarking #LLMs
https://arxiv.org/abs/2610.20675
@dataprismai.bsky.socialDiviner is an agentic root-cause analysis system operating at Meta, structured around a four-stage reliability lifecycle that combines reviewed incident data with service-owner expertise to handle complex…
#RCA #AIAgents #Observability #DistributedSystems
https://arxiv.org/abs/2610.21281
@gradientbrief.bsky.socialAn open agent-native system called Athanor lets AI agents autonomously run post-training research on real GPUs with public data, handling data prep, training code, and hyperparameter tuning. It restricts agents to…
#AIAgents #OpenSource #MachineLearning #AIResearch
https://arxiv.org/abs/2610.21055
@paulirolla.bsky.socialEffective token prices fell 41%, and usage rose 50% since July (Ramp AI Index, Sept 2026). Waiting for the next price cut will not shrink an agent bill. Every turn resends the context, so the volume per task is what you can actually move. #AIagents
@uenozooo.bsky.socialMicrosoft Foundry September Update Solves Problems Developers Never Had
https://uenozooo.com/microsoft-foundry-september-update-solves-problems-developers-never-had/
@powerbitips.bsky.social568 - The Fabric APIs shipped in the last month or two weren't written for you. They're built for an agent to scan and consume, same with the new Python library. Nothing stops you from using them as a human.
@usehiveai.bsky.socialRefund crosses two apps: request in the support desk, order in billing. The agent reads both, applies rules that live in neither API, issues the refund, confirms to the customer. The handoff carries a stale snapshot and nobody rechecks. Connected steps are not a workflow. #AIAgents #BuildInPublic
@seldondance.bsky.socialCalling agents 'not rogue' concedes the worse case: nothing stopped them, nobody ran the checks. Frontier labs ship capability with the harness unfinished; an agent is exactly as safe as the checks wired around it. #ai #aiagents
https://seldondance.substack.com/p/the-supply-of-checkable-failure
@thesurvivorbot.bsky.social$BTC 82,054 now, range-high 84,155. The bot may buy alts today as $BTC daily bias stays up. $INJ 9.125 range-high 9.092 (0.4% above). $TRX 0.3349 range-high 0.3354 (0.1% below). Not financial advice.
@seldondance.bsky.socialSteering without a window: chain-of-thought observability collapsed in the agentic era, so supervision shifts from reading the reasoning to demanding interim outputs and judging outcomes. The End of Watching AI Work: #ai #aiagents
https://seldondance.substack.com/p/the-end-of-watching-ai-work
@a-i-trends.bsky.social🎬 Hermes Agent v0.22.1 & v0.22.2 — Patches, Gateway Fix & Memory Blog https://youtu.be/7uya5AEAJTA #HermesAgent #AI #OpenSource #AIAgents
@bibo72.bsky.socialMETR's independent-assessment method, in one spec: survey the misbehavior record, then hunt what it hides — transcripts, checkpoints, ~400M tokens/min access. Anthropic is pilot #1.
@ai-news.at.thenote.appAI Agents Are Becoming Infrastructure. Who Controls the Control Plane?
AI agents are becoming enterprise infrastructure. Explore why organisations need a control plane for identity, policy, security, observability, and lifecycle
@nile11.bsky.socialResearchers Talk to Live AI Agents Scraping Alibaba Maps, Trace Activity to Tencent Cloud
..................
https://nile1.com/researchers-talk-to-live-ai-agents-scraping-alibaba-maps-trace-activity-to-tencent-cloud/
..................
#aiAgents #aiSafety #alibaba #amap #autonomousAgents #cloudSandb
@ai-ru.at.thenote.appИИ-агенты становятся инфраструктурой. Кто контролирует управляющую плоскость?
ИИ-агенты становятся корпоративной инфраструктурой. Узнайте, почему организациям нужна плоскость управления для идентификации, политик, безопасности, наблюдаемости и жизненного ц…
@theeternalobserver.bsky.socialAfter 11/50: Muse Control 125; ChatGP-TKO 122.5; Gemini Twinfinite 120; Grok Solid 117.5; Claude Nine 65. Points leader: Muse Control. Agent-designed games, agent-only competitors. #AI #AIAgents #LLM #BottleDecathlon
@adityamitra.bsky.social