enterprise data management

'Welcome to the AGI era': OpenAI launches GPT-6 Astra

OpenAI has ushered in a new era with the release of GPT-6 Astra, a model poised to redefine how we interact with technology and potentially mark the onset of artificial general intelligence (AGI). Astra moves beyond traditional chatbots, enabling users to direct AI through voice commands to navigate software, automate workflows, and produce finished documents – effectively eliminating the need for constant mouse clicks or keyboard input.

Microsoft AI’s MAI-Transcribe-2 undercuts OpenAI, Google and ElevenLabs on price and speed

Microsoft AI has significantly disrupted the speech recognition landscape with the release of MAI-Transcribe-2, undercutting OpenAI, Google, and ElevenLabs on both price and speed. Priced at just 10 cents per hour, this represents a remarkable 72% reduction from the initial model's cost. Offering features like speaker diarization, word-level timestamps, and code switching—typically premium capabilities—for this price, MAI-Transcribe-2 positions itself as a compelling solution for enterprises processing substantial audio volumes. For those interested in exploring this evolving market, “Meta prices Muse Voice Transcribe at $0.

Enterprises put non-Nvidia chips 14 points ahead of Nvidia's next-gen GPUs on their evaluation lists

Recent VentureBeat research reveals a significant shift in enterprise AI accelerator strategy. While Nvidia remains dominant in production environments, a striking 39.4% of organizations are now actively evaluating non-Nvidia alternatives like AWS Trainium and Google TPUs – a 14-point increase over Nvidia's next-gen GPUs. This indicates a move toward greater optionality and workload-level scrutiny, with organizations prioritizing integration, performance, and cost-effectiveness. Enterprises are increasingly seeking control over their AI infrastructure, a trend underscored by growing interest in open-source components.

Stolen Claude session cookies can reach corporate Gmail through grants no IT admin can revoke

A recent campaign exploiting common infostealer malware has exposed a critical vulnerability: stolen Claude session cookies can grant access to corporate Gmail accounts, bypassing traditional security measures. Attackers replay stolen cookies to access paid accounts, sidestepping two-factor authentication and single sign-on (SSO) limitations. Anthropic has responded by signing out affected accounts and refunding charges, but the potential exposure of conversation history and connected applications remains a concern.

Frontier models can recover up to 65% of facts they can't directly recall — just by thinking longer

Recent research from Google and Technion reveals a surprising truth about large language models (LLMs): they often *possess* the knowledge needed to answer questions, but struggle to retrieve it. Frontier models like GPT-5 and Gemini-3 encode up to 98% of tested facts, yet fail to directly recall 26-34% without additional processing. This highlights a critical shift – focusing on improving *access* to existing knowledge through inference-time computation, rather than solely scaling models, can unlock significant gains in factual accuracy.

Closing an Azure OpenAI assistant's retrieval gap didn't take a new identity platform. It took one filter and a narrower assistant.

Recent incidents highlight a critical vulnerability in many Azure OpenAI deployments: retrieval-time entitlement failures. Egiziago Cioffi, CEO of SynSphere Italia, discovered this firsthand when his email assistant, despite passing all evaluations, inadvertently exposed SharePoint content to users lacking proper access. Cioffi’s solution? A simple filter that checks user permissions before content reaches the model. This underscores a vital point: current evaluations often miss these boundary breaches.

Anthropic's Claude Fable 5.1 and Mythos 5.1 arrive with a 75% cost reduction for Fable cache reads

Anthropic has released Claude Fable 5.1 and Claude Mythos 5.1, the latest iterations of its powerful large language models, alongside a significant 75% cost reduction for Fable cache reads. These models prioritize sustained problem-solving, demonstrating substantial improvements on benchmarks like Terminal-Bench and AutomationBench. Crucially, Anthropic is also introducing Enterprise Frontier Safeguards (EFS), allowing organizations to retain monitoring data within their own infrastructure. This release addresses evolving enterprise needs for capable, economical, and governable AI agents—a shift underscored by recent cybersecurity evaluations.

Your files stay put: Perplexity’s hybrid AI keeps confidential data off the cloud

Perplexity today introduces hybrid AI compute, a transformative system designed to keep your confidential data secure. Computer, Perplexity’s agentic platform, now intelligently splits tasks between cloud-based and locally-run AI models on Apple silicon Macs, ensuring sensitive information never leaves your device. This innovative approach combines the power of frontier models with the privacy of on-device processing, a critical advancement for industries handling sensitive data. Explore this new capability today and discover how Perplexity is redefining data security and productivity.

OpenClaw 2.0 is here, ushering in the era of 'multiplayer' AI coding: What it means for enterprises

OpenClaw 2.0 is here, marking a significant shift toward enterprise-ready AI coding. Building on the viral momentum of earlier versions, this update transforms OpenClaw from a personal agent harness into a collaborative platform designed for teams and shared infrastructure. Key additions include a rebuilt browser interface, shared cloud sessions, and enhanced security features like role-based permissions and auditing. For organizations, OpenClaw 2.0 envisions agents as a shared operational layer, not just individual developer tools—a concept DoorDash recently explored with its Flux platform.

AI agents that pass authentication can still drift, expose data, or get memory-poisoned

Securing AI agents requires a shift in perspective. While gateways are often the initial defense, they're frequently deployed before foundational identity and attribution layers are in place, creating a significant vulnerability. Recent events, like the CISA advisory regarding a LiteLLM flaw, highlight this risk. Prioritize establishing agent inventory, distinct identities, and task-scoped credentials *before* relying on runtime enforcement. Start with the basics – identifying and naming your agents – to build a robust security foundation.

AI agents need their own identity before they need a gateway

Enterprise AI has entered a new era, moving beyond simple assistants to autonomous agents capable of complex workflows. This shift introduces a fundamental security challenge: authentication confirms identity, but it doesn't guarantee ongoing trust. Traditional security controls offer limited visibility into an agent’s actions after authentication, creating new runtime risks like goal drift and memory poisoning. To address this, organizations must embrace runtime trust – continuously validating AI behavior and ensuring alignment with organizational policy.

Cohere Parse 5 loses the benchmark on points. It wins on cost per page.

Enterprises seeking to integrate PDFs, slides, and scanned documents into AI pipelines often encounter a critical bottleneck: balancing accuracy with cost. Cohere’s Parse 5 addresses this challenge, prioritizing price-to-performance over raw accuracy. While benchmark results show Parse 5 trailing larger models like GPT-5.5, it delivers a compelling value proposition, costing just $1.50 per 1,000 pages. This strategic approach makes enterprise-scale document parsing more economical, a crucial step in realizing the potential of agentic AI, as highlighted in our recent article on agentic AI security.

The three layers of agentic AI security: A defense-in-depth architecture for autonomous agents

Autonomous agents, capable of independent reasoning and action, introduce unique security risks that traditional application controls can’t address. Nutanix proposes a defense-in-depth architecture, structured across three critical layers: infrastructure, network, and control plane. This approach, detailed by Nutanix's Oscar Wahlberg, establishes a layered security posture, ensuring robust protection against everything from unauthorized access to runaway agent behavior.

Meta researchers taught an 8B AI model to match Claude Opus 4.5 — without the frontier price tag

Researchers at Meta AI and the University of Illinois Urbana–Champaign have developed EvoHarness-RL, a framework that significantly enhances AI agent performance in complex, long-horizon tasks. This innovation teaches AI models, like Qwen3-8B, to intelligently manage their runtime environment, improving efficiency and accuracy—even rivaling larger, costly models like Claude Opus 4.5. By consolidating agent support systems into a unified Belief, Progress, and Experience workspace, EvoHarness-RL offers a path toward more adaptable and cost-effective AI solutions, as explored further in "Enterprise AI's real risk isn't autonomous agents.

When agents act on their own, governance has to live in the data layer

Salesforce just put its entire CRM inside Claude — and says you’ll never need its app again

Salesforce and Anthropic are redefining enterprise software with Claudeforce, a new plugin that brings the entire CRM platform directly into Claude. This innovative integration, available to select customers today, empowers sellers to query, update, and act on live CRM data without ever opening Salesforce itself—potentially eliminating thousands of clicks per morning. Salesforce envisions a future where the UI *is* the AI, allowing for dynamic app creation and personalized workflows.

Rethinking Data Visualisation: A UX Approach To Dashboards That Actually Drives Decisions

Data visualization often falls short of driving meaningful decisions, hampered by a disconnect between data and design. Rethinking Data Visualisation explores a transformative approach: applying structured UX thinking to dashboards and data presentations. Meriem Benhabiles guides you through a process, from initial questioning to impactful insight delivery. This isn't about aesthetics; it's about ensuring data truly informs action. For a deeper dive into related AI challenges, explore "How Does a RAG Reranker Really Work?" and discover enterprise document intelligence.

Orchestration is the new challenge for CX in the age of AI agents

The rise of AI agents presents a new challenge for customer experience: orchestration. As enterprises rapidly deploy AI across channels, many are struggling to integrate these tools with legacy systems, creating fragmented customer journeys and overburdened human agents. Tata Communications’ Gaurav Anand explains that the shift is moving away from simple automation toward intelligent orchestration—connecting tasks and delivering end-to-end outcomes with a shared understanding of the customer. Discover how this approach, underpinned by a common enterprise ontology, can transform CX.

Anthropic’s new Claude Tag update lets its Slack agent read the full conversation — and jump in unprompted

Anthropic’s latest Claude Tag update marks a pivotal shift in enterprise AI. Now, Claude's Slack agent reads entire conversations, proactively offering assistance—sometimes unprompted—a move Anthropic calls "multiplayer AI." This represents a transition from individual AI tools to collaborative agents embedded within teams, streamlining workflows and boosting productivity. According to Anthropic, this change improves decision-making by roughly 30%.

Enterprise AI agents are only as reliable as the messiest documents behind them

Enterprise AI's potential is often hampered by the disorganized data underpinning it. While context engineering—connecting systems, generating embeddings, and building retrieval pipelines—works for isolated assistants, it treats enterprise knowledge as application-specific, leading to inconsistency and duplicated effort. As AI deployments expand, managing enterprise knowledge itself becomes paramount. A shared enterprise knowledge platform, akin to an enterprise data platform, offers a solution, organizing knowledge into layers for preservation, normalization, integration, and optimized serving—a foundation for reliable, scalable AI.

Enterprises winning with AI agents are limiting how much the agents can do alone

Enterprises are discovering a critical truth about AI agents: unrestrained autonomy isn't synonymous with superior performance. While the initial focus was on maximizing agent independence, current deployments reveal that controlled, narrowly-scoped agents, coupled with strategic human checkpoints, are proving far more sustainable. Gartner forecasts that over 40% of agentic AI projects won't reach 2028, highlighting a widening gap between capability and responsible AI maturity.

Nvidia finds that simple linear math can replace costly AI model handoffs

Nvidia researchers have uncovered a significant inefficiency in agentic AI systems: the costly recomputation of conversation history when switching between models. To address this, they’ve introduced a cross-model KV cache transfer technique utilizing simple linear math, dramatically reducing compute costs and latency. Experiments reveal this method can be 2.7 to 25 times faster than traditional recomputation, retaining up to 98% of accuracy. This innovation paves the way for more efficient, long-horizon, multi-LLM workflows, as explored further in our article, "PagedAttention vs.

Slack wants to drag AI coding out of the terminal and into the group chat

Slack is redefining AI coding, moving it from isolated terminals into collaborative group chats with Slack Code. This new product embeds AI coding agents – including Anthropic's Claude Code, Cognition's Devin, and others – directly into dedicated Slack channels, enabling teams to collectively review, steer, and ship software. Slack is betting that multiplayer AI, where everyone can see and contribute, will lead to higher quality and faster development, effectively removing code as the primary bottleneck.

One in five enterprises can't stop a runaway AI agent's spending in real time

Enterprise adoption of AI agents is revealing a critical shift: organizations are increasingly deploying multiple orchestration platforms—averaging three—to mitigate vendor risk and retain control. This trend, driven by concerns around security, permissions, and visibility, sees Microsoft AI Foundry/Copilot Studio leading usage, with Anthropic's Claude Platform gaining significant consideration. Notably, one in five enterprises still lacks real-time control over agent spending, highlighting the need for robust oversight as AI deployments evolve. Learn more about this emerging landscape with VentureBeat's coverage of Serval’s AI agent, Catalyst.