AIStarMap

News & Updates

June 2, 2026

Anthropic expands Project Glasswing to 150+ organizations, discovering over 10,000 high-severity vulnerabilities

Anthropic has expanded Project Glasswing from 50 to 150+ organizations (15+ countries), covering critical infrastructure areas such as power/water/health/communications, etc., and partners have discovered more than 10,000 high-risk/critical vulnerabilities.
Source: Anthropic official announcement
August 2, 2026

Amazon completes $50 billion investment in OpenAI, 10-Q discloses conversion of preferred shares to common shares

Amazon’s 10-Q quarterly report disclosed that it has completed a full investment of US$50 billion in OpenAI. After OpenAI’s IPO, the Series C preferred shares were converted into ordinary shares, and the binding of cloud computing power and cutting-edge models entered the substantial implementation stage.
Source: user feed
March 31, 2026

Zendesk 2026 Spring Edition: AI Agent autonomous customer service is online, and the "perception-decision-execution" closed loop of the work order system is formed

Zendesk 2026 spring version adds AI Agent autonomous customer service agent, enhanced intent recognition and workflow automation engine, deeply embedding LLM capabilities into the work order life cycle.
Source: Zendesk Official Release

Salesforce Spring '26: Einstein Copilot is upgraded to an autonomous agent platform, and Service Agent and SDR enter the automatic execution stage

Salesforce Spring '26 upgraded Einstein Copilot to an autonomous agent platform, launched Einstein Service Agent autonomous customer service and Einstein SDR automatic outbound calls and lead cultivation, and deeply integrated Data Cloud real-time data flow.
Source: Salesforce official release

Einstein Spring '26: Einstein Copilot agent goes online, CRM native AI moves from prediction to automated execution

Einstein 2026 Spring Release adds the Einstein Copilot agent, enhanced predictive analysis and automated recommendations, pushing CRM native AI from predictive analysis to automated execution.
Source: Salesforce official release
June 10, 2026

Writer releases WRITER Agent update: generated PPTX can be exported to Google Slides, and the enterprise Agent collaboration link is opened again

Writer official What's New update: PPTX deliverables generated by WRITER Agent can be exported to Google Slides, strengthening collaboration links for corporate documents and presentations.
Source: Writer official release
November 30, 2023

Whisper large-v3: the "ceiling" version of open source speech recognition, the accumulation of 99+ language transcription and 680,000 hours of weakly supervised training

Whisper large-v3 is currently the largest public Whisper model version, reaching the previous highest level in multi-language transcription and translation quality, and inheriting the iterative accumulation of 680,000 hours of weakly supervised training in 2022.
Source: OpenAI official release
April 30, 2026

Tongyi Qianwen Qwen3 released: Alibaba Cloud's flagship model leads the industry in many capabilities, and "cloud + terminal" collaboration enters a new stage

Tongyi Qianwen's latest flagship model Qwen3 was released, with many capabilities reaching industry-leading levels, continuing the improvements in code and mathematical reasoning of Qwen2.5, and deeply integrating with the Alibaba Cloud ecosystem.
Source: Alibaba Cloud official release

Intercom 2026 Q2: Fin AI Agent introduces workflow execution, and AI customer service for outcome billing goes a step further

Intercom 2026 Q2 introduces Fin Procedures workflow execution capabilities to Fin AI Agent, strengthens Copilot and Insights analysis, and continues the AI ​​customer service model billed by outcome.
Source: Intercom official release
July 31, 2026

Sentry 2026 platform update: From error tracking to full-stack observability, Seer AI turns "post-event troubleshooting" into "pre-emptive prompts"

The current version of Sentry covers error monitoring, distributed tracing, performance analysis, Session Replay and Logs/Metrics, and uses Seer AI to introduce AI into observability analysis, serving 150K+ organizations.
Source: Sentry official release

Semrush 2026.07: AI intelligent content suggestions and automated reports are launched, and the digital marketing platform enters the "AI full process" stage

Semrush 2026.07 adds new AI-driven intelligent content suggestions and automated report generation, superimposes previous content marketing scores and competitive product advertising insights, and accelerates the AI-based marketing platform covering 140+ national databases.
Source: Semrush official release

LM Studio 1.0: Local large model running enters a "desktop-level" experience, and the threshold for open source LLM on consumer-grade hardware is lowered again

LM Studio 1.0 continues its positioning as a desktop application, allowing users to download, run and manage open source large language models locally without coding and cloud dependence. It is the entry-level benchmark for local AI inference.
Source: LM Studio official release

The HubSpot platform continues to iterate: AI content agency and intelligent prediction go deep into CRM, and the "automation" of sales and marketing moves toward "autonomy"

HubSpot will continue to iterate AI functions and platform integration in 2026, and AI content agency, intelligent prediction and enhanced automated workflow will gradually penetrate into the entire sales, marketing and service process.
Source: HubSpot official release

Figma 2026-07 update: AI-assisted design and Dev Mode collaboration are further polished, and the "design delivery" link continues to be automated

Figma 2026-07 monthly update continues to optimize the AI-assisted design function and Dev Mode collaboration experience, taking over the official launch of Figma AI in 2026-04 and the AI ​​assistant preview of Config 2025.
Source: Figma official release

ElevenLabs voice ecosystem is further advanced: Scribe v2 and ElevenAgents Expressive Mode upgrade "speaking" to "expressing"

ElevenLabs continues to expand the voice ecosystem. Scribe v2 speech recognition and ElevenAgents’ expression capabilities have been enhanced. Together with Music v2 and Dubbing v2, the three-layer product structure (creation/conversation/API) has been simultaneously strengthened.
Source: ElevenLabs official release
July 14, 2026

OpenAI API 1.0 public beta release: the multi-modal capability matrix is ​​formed, and the "configuration" delivery of the developer platform goes one step further

OpenAI API entered the 1.0 public beta stage on July 14, 2026, covering text, images, speech and embedded vectors with a multi-modal capability matrix, providing developers with model gradients from flagship high accuracy to lightweight and low-cost.
Source: OpenAI official release
May 14, 2026

LlamaIndex 0.14.22: RAG data framework continues to iterate, and the ecosystem of connectors and retrieval components is strengthened

LlamaIndex 0.14.22 was released on 2026-05-14, continuing the main line of data loading, indexing, retrieval and agent orchestration, and continuing to improve the integration with various models, vector libraries and data sources.
Source: LlamaIndex official release
June 30, 2026

Jina AI ReaderLM-v2 enhanced release: the conversion quality from HTML to Markdown/JSON has been improved, and the Agent retrieval base has been upgraded.

Jina AI enhances the HTML to Markdown/JSON conversion quality and expands search capabilities in Reader, and ReaderLM-v2 allows web content to become input to LLM and Agent more reliably.
Source: Jina AI official release

Grafana 11.5 released: AI alarm suggestions and natural language queries are in place, and the open source observable stack enters the "AI native" era

Grafana 11.5 was released on 2026-06-30, adding AI-driven alarm suggestions, Grafana LLM Plugin enhancements and natural language query dashboard upgrades, continuing the main line of AI observability since 11.4.
Source: Grafana Labs official release
September 30, 2025

Codex 1.0 GA: GPT-5 driven cloud parallel programming Agent, PR delivery model is officially available

The official version of Codex 1.0 has entered GA from research preview, supporting cloud parallel tasks, CLI and IDE access. The bottom layer has been upgraded to the GPT-5 series model for coding optimization, and the results are delivered in the form of PR.
Source: OpenAI official release