Agentic AI

Twitter

Facebook

Copy Link

Key Facts

Year2024

Key PlayersOpenAI, Anthropic, LangChain

Top News

NVIDIA Nemotron Achieves Benchmark-Leading Performance With LangChain Deep Agents Blueprint

NVIDIA Nemotron 3 Ultra delivers top accuracy among open models on LangChain's Deep Agents benchmark, achieving business task parity with leading closed models at 10x lower inference cost. The NemoClaw for LangChain Deep Agents blueprint combines an open-weight model, agent harness, and secure runtime to help enterprises build, evaluate, and deploy advanced AI agents they can fully customize and control.

Technology2 sources

NVIDIA Nemotron Achieves Benchmark-Leading Performance With LangChain Deep Agents Blueprint

Technology2 sources

GitHub Copilot produces harmful content via workflow-level jailbreak despite chat refusals

Researchers at the Alan Turing Institute discovered that GitHub Copilot and other AI coding agents refuse dangerous requests in chat but produce the same harmful content when embedded in ordinary software development workflows. The study tested four models across 816 attempts, revealing a critical gap in how the industry measures AI safety.

Technology3 sources

GitHub Copilot produces harmful content via workflow-level jailbreak despite chat refusals

Technology3 sources

Claude Fable 5 dominates Google's revamped Android AI dev benchmark with 84.5% accuracy

Google has overhauled Android Bench, its AI model benchmark for Android development, revealing Claude Fable 5 leads with 84.5% accuracy. The updated framework adds eight new models and switches to Harbor for easier testing. Despite the improvements, Google's own Gemini models lag behind competitors in Android coding tasks.

Technology2 sources

Claude Fable 5 dominates Google's revamped Android AI dev benchmark with 84.5% accuracy

Technology2 sources

AI Security Crisis Unfolds as 69% of Enterprises Share Agent Credentials Across Deployments

New research from VentureBeat exposes a critical vulnerability in enterprise AI deployments: 69% of organizations share credentials across AI agents, creating cascading security risks. With 54% already experiencing AI agent security incidents and autonomous cyberattacks moving in 27 seconds, the findings explain why Palo Alto Networks, CrowdStrike, and Cisco have collectively invested over $22 billion in AI security solutions targeting this exact weakness.

Technology6 sources

AI Security Crisis Unfolds as 69% of Enterprises Share Agent Credentials Across Deployments

Technology6 sources

Varonis Systems extends AI security coverage to Cursor AI coding tool via Atlas platform

Varonis Systems announced support for Cursor, the AI-native coding tool, through its Varonis Atlas platform. The integration delivers runtime enforcement, threat detection, and session forensics to protect sensitive data as AI agents access source code, credentials, and API keys across development environments.

Technology2 sources

Varonis Systems extends AI security coverage to Cursor AI coding tool via Atlas platform

Technology2 sources

Claude Cowork earns its place: AI tool tackles email chaos and saves hours on tedious tasks

Multiple journalists tested Anthropic's Claude Cowork on real-world tasks from managing Gmail inbox overload to reviewing contracts and organizing files. The AI-powered desktop application demonstrated significant time-saving potential, though privacy concerns remain. One user unsubscribed from 21 newsletters in five minutes, while another sifted through 7,000 emails to find specific PR pitches.

Technology3 sources

Claude Cowork earns its place: AI tool tackles email chaos and saves hours on tedious tasks

Technology3 sources

Slackbot now pulls CRM data, generates charts, and sends DocuSigns directly from chat

Five years after Salesforce acquired Slack for $27.7 billion, the two platforms are finally operating as one integrated system. Slackbot, the AI assistant built into every Slack workspace, can now access the entire Salesforce ecosystem—including CRM records, Tableau analytics, Data 360 profiles, and third-party apps—through simple conversational prompts, eliminating the need to switch between applications.

Technology2 sources

Slackbot now pulls CRM data, generates charts, and sends DocuSigns directly from chat

Technology2 sources

Amazon's Project Moonraker aims to turn Alexa into an AI agent, with GPU costs exceeding $100M

Amazon is developing Project Moonraker, an ambitious upgrade to transform Alexa into a fully agentic AI assistant capable of handling complex multi-step tasks. Internal documents reveal projected GPU costs exceeding $100 million in 2026, making it the highest-cost initiative in the Alexa+ overhaul. The project aims to compete with AI agents from OpenAI, Google, and Anthropic.

Technology3 sources

Amazon's Project Moonraker aims to turn Alexa into an AI agent, with GPU costs exceeding $100M

Technology3 sources

Perplexity builds Teammate AI coding tool to challenge Cursor, Claude Code and OpenAI

Perplexity, known for its AI search engine, has quietly developed an internal AI coding tool called Teammate. Engineers have used it since May for bug investigation and service monitoring. If launched publicly, the $20bn startup would enter direct competition with Cursor, Anthropic's Claude Code, and OpenAI in the lucrative AI coding market.

Technology3 sources

Perplexity builds Teammate AI coding tool to challenge Cursor, Claude Code and OpenAI

Technology3 sources

Former GitHub CEO launches Entire to solve AI coding agents' strain on centralized platforms

Thomas Dohmke, former GitHub CEO, has launched Entire with a distributed Git network designed to handle the explosive demand from AI coding agents. The platform allows developers to mirror GitHub repos across regional nodes, claiming to sustain 2.1 million pushes per hour. Built on a $60 million seed round, Entire addresses the infrastructure stress that's been overwhelming centralized platforms like GitHub.

Technology4 sources

Former GitHub CEO launches Entire to solve AI coding agents' strain on centralized platforms

Technology4 sources

Box Survey Reveals Enterprise AI Leaders Outperform Peers Through Content Access and Governance

A new Box survey of 1,640 IT decision-makers shows enterprise AI adoption surged from 8% to 64% in one year. Leading companies achieve 25%+ ROI by integrating AI agents with trusted content and formal governance, while half of all organizations report AI-related data exposure incidents. Box CEO Aaron Levie argues both frontier and specialized AI models will drive spending growth for years.

Technology2 sources

Box Survey Reveals Enterprise AI Leaders Outperform Peers Through Content Access and Governance

Technology2 sources

HalluSquatting attack exploits AI hallucinations to build botnets by tricking coding assistants

Security researchers from Tel Aviv University revealed HalluSquatting, an AI security threat that exploits AI hallucinations to trick AI agents into executing malicious code. The attack targets popular AI coding assistants like Cursor, GitHub Copilot, and Gemini CLI, with hallucination rates reaching 85% for repositories and 100% for trending skills. Attackers register fake resource names predicted by large language models, enabling AI-enabled botnets at unprecedented scale.

Technology5 sources

HalluSquatting attack exploits AI hallucinations to build botnets by tricking coding assistants

Technology5 sources

Storage Technology Takes Center Stage as Agentic AI Demands Explode Beyond GPU Capabilities

As AI evolves from simple chatbots to autonomous agentic systems, the real bottleneck isn't GPU processing power—it's data storage. With million-token context windows generating petabytes of data, storage technology is being redefined as the intelligence layer where raw data becomes actionable intelligence. Industry leaders from Solidigm to Western Digital are racing to address what could be AI's trillion-dollar storage challenge.

Technology3 sources

Storage Technology Takes Center Stage as Agentic AI Demands Explode Beyond GPU Capabilities

Technology3 sources

AI hacking skills outpace testing benchmarks as US federal agencies face August 1 deadline

Frontier AI models now develop hacking capabilities faster than the tests designed to measure them. With US federal agencies facing an August 1 deadline to establish benchmarking standards, existing evaluations already look obsolete. Industry leaders warn that static tests fail to capture what these systems can do in real-world environments, leaving regulators and security teams unable to assess whether AI models pose genuine cybersecurity threats.

Policy3 sources

AI hacking skills outpace testing benchmarks as US federal agencies face August 1 deadline

Policy3 sources

Nvidia Vera CPU targets AI agent workloads with focus on single-threaded performance

Nvidia frames its Vera CPU as the first 'max single-threaded CPU at scale,' designed specifically for AI agent workloads. Perplexity confirms adoption, reporting 1.5x faster performance in agentic coding tasks compared to traditional x86 processors, as Nvidia expects $20 billion in Vera sales this fiscal year.

Technology5 sources

Nvidia Vera CPU targets AI agent workloads with focus on single-threaded performance

Technology5 sources

Prev 123 4…87 Next