-
Money · SiliconANGLE · 19 Oct
SiliconANGLE reports that Town raised $108 million in a Series B at a $1 billion valuation. Town said it will use the money to add more integrations for its AI work assistants, which launched in June.
Why it matters. It points to more integrations for AI work assistants that companies use in daily work.
Agents at workAssistantsCompaniesPermalinkRead at siliconangle.com
-
Launch · SiliconANGLE · 19 Oct
SiliconANGLE reports that Motive Technologies launched Agents, autonomous AI agents for fleet operations that predict problems and act without constant human supervision. Motive says existing customers can deploy them in minutes to handle compliance, maintenance and safety tasks.
Why it matters. It shows agent workflows moving from alerts to autonomous actions in real operations, with guided setup for compliance, maintenance and safety.
SafetyAgents at workCompaniesPermalinkRead at siliconangle.com
-
Analysis · TechCrunch · 19 Oct
TechCrunch reports Bloomberg said Anthropic and Nscale may announce another $45 billion compute deal. It also says talks are ongoing, details may change, and both companies are pursuing IPOs.
Why it matters. Another large Anthropic compute deal could shape how teams plan infrastructure around Anthropic.
InfrastructureCompaniesPermalinkRead at techcrunch.com
-
Launch · TechCrunch · 19 Oct
TechCrunch reports Nvidia launched open-weight models for always-on agents and edge computing, including the 120-billion-parameter Nemotron 3.5 Super VL and two Bolt models for local use. Nvidia says Super VL and Bolt 74B-A9B ship now, while Bolt E4B is due later this month.
Why it matters. This adds open models for enterprise agents on Nvidia infrastructure and for local assistants on AI PCs.
ModelsAssistantsInfrastructurePermalinkRead at techcrunch.com
-
Money · WIRED · 19 Oct
WIRED reports that court filings tie Character.AI’s bankruptcy to a Disney deal covering most staff and a bid for a broad license to its technology. Filings say the structure leaves Character.AI responsible for ongoing lawsuits and regulatory matters.
Why it matters. For AI companies, it shows how a buyer can hire employees and license technology while the startup remains responsible for lawsuits and regulatory matters.
SafetyCompaniesPermalinkRead at wired.com
-
Analysis · Amazon Web Services · 19 Oct
AWS reports that bi-directional KV cache transfer in vLLM lets prefill and decode nodes reuse context across turns. In 1P1D benchmarks on Amazon EC2 P5en, it reduced TTFT by as much as 1.6x with Qwen3-32B and 3.3x with zai-org/GLM-5-FP8.
Why it matters. For multi-turn agents, this can reduce repeated prefill computation and improve TTFT as conversation history grows.
Agents at workInfrastructurePermalinkRead at aws.amazon.com
-
Research · Hugging Face · 14 Oct
Hugging Face reports that OpenMobile-2 is an open mobile agent that uses simulated apps, supervised fine-tuning and RL data, plus GUI actions and app-native tools. The authors say its playground spans 110+ apps, and OpenMobile-2 scored 79.9 on AndroidWorld and 50.4 on MobileWorld.
Why it matters. This gives mobile agent builders open apps, data, and benchmarks for training and testing across GUI actions and app-native tools.
ModelsDevicesInfrastructurePermalinkRead at huggingface.co
-
Launch · The Decoder · 19 Oct
THE DECODER reports MiniMax released M3.1 Flash, a reasoning model that scores 42 on the Artificial Analysis Intelligence Index at about $0.22 per task. It handles up to one million tokens, accepts text and images, and MiniMax plans an open-weight release on October 23.
Why it matters. For agent builders, this adds a lower-cost model with one million tokens of context, text and image input, and planned open weights.
ModelsInfrastructureCompaniesPermalinkRead at the-decoder.com
-
Launch · TechCrunch · 19 Oct
TechCrunch reports Laplace Intelligence is building automated red-teaming for robots, autonomous vehicles, and other physical AI in simulation before deployment. CEO Howie Sun said the startup uses AI agents to find weaknesses and has customers including Bot Auto and a Japanese automotive manufacturer.
Why it matters. This adds a way to probe edge cases in physical AI with AI agents before deployment.
SafetyInfrastructureCompaniesPermalinkRead at techcrunch.com
-
Launch · SiliconANGLE · 19 Oct
SiliconANGLE reports that Pax8 launched 8os so IT service provider partners can govern AI agents across customer businesses, alongside 8bot, a managed AI coworker for employees. Pax8 said limited early access opens this week, with general availability planned for early next year.
Why it matters. This gives IT partners a way to deploy and supervise AI agents and a managed AI coworker for customers through the Pax8 marketplace.
Agents at workInfrastructureCompaniesPermalinkRead at siliconangle.com
-
Launch · TechCrunch · 19 Oct
TechCrunch reports Microsoft launched Guardian Agents and new Defender tools to monitor enterprise AI agents, detect up to 60 agent types, and curb data or system access. Microsoft says frontier models cannot be trusted, and Nadella has urged companies to separate model supply from control over agents.
Why it matters. For teams deploying agents, the release adds ways to watch agent behavior, limit permissions, and require human approval for risky decisions.
SafetyInfrastructureCompaniesPermalinkRead at techcrunch.com
-
Launch · Cloudflare Blog · 19 Oct
Cloudflare Blog reports Cloudflare previewed Clairic, a System One extraction model for Workers AI. The post says it returns typed answers from text and up to four optional images in one pass, with confidence scores and no output tokens.
Why it matters. This gives people running agents on Workers AI a typed extraction model that handles text and images in one pass without output tokens.
ModelsInfrastructurePermalinkRead at blog.cloudflare.com