Firefly 的 Generate Music、Generate Speech 和 Generate Sound Effects 已全面开放；Adobe 称其自有音频模型生成的音乐与声音可安全用于商业制作。
Generate Music, Generate Speech, and Generate Sound Effects are now broadly available. Adobe describes output from its own audio models as commercially safe for finished work.
官方 Product UGC 工作流接收人物图与商品图，可加入商品信息、语气或完整脚本，生成 4–15 秒竖版带声推广视频。
The official Product UGC recipe combines a creator image and a product image with an optional brief or script to produce a 4–15 second vertical promo with audio.
H3 可同时理解文本、图片、视频与音频，生成最长 15 秒、最高 2K、带 32 kHz 立体声的视频；基础模型已开源。
H3 accepts text, image, video, and audio context and generates up to 15 seconds of 2K video with 32 kHz stereo audio. The base model is open.
UCP 与 Catalog API 现可由开发者自助使用，覆盖商品搜索、购物车和结账，并加入图片与多模态搜索。
UCP and Catalog API are now self-serve for developers, spanning product discovery, carts, and checkout with image and multimodal search.
OJO 将产品目标、研究、PRD、UI、交互原型和开发交付放进同一工作区；当前为 private beta。官方未标注发布时间，本次按收录日期入站。
OJO connects product goals, research, PRDs, UI, interactive prototypes, and development handoff in one workspace. It is in private beta; the official source does not state a publication date, so this entry uses its filing date.
Lovart 可读取提示词、参考文件和 Brand Kit，在无限画布中生成并编辑商品图、社交素材、营销套件和品牌视频；本条为此前遗漏的工具补录。
Lovart uses prompts, reference files, and Brand Kits to create and edit product shots, social assets, campaign kits, and brand videos on an infinite canvas. This is a catch-up entry for a previously missed tool.
OpenAI 案例显示，Stampli 将估算的 243 小时生产工作压缩至约 77 小时；所有面向客户的内容仍由人工审核。
OpenAI reports that Stampli reduced an estimated 243 production hours to about 77, with human review for all customer-facing work.
xAI 称 Grok Build 已面向所有套餐开放，可在网页与移动端生成并发布应用、游戏、网站和看板，并支持导出到 GitHub。
xAI says Grok Build is available on every plan across web and mobile, with publishing and GitHub export.
Shipt 在 ChatGPT 与 Claude 中推出美国购物体验，可根据自然语言生成个性化购物车，再跳转至 Shipt 完成结账。
Shipt launched U.S. shopping experiences in ChatGPT and Claude that build personalized carts and hand off to Shipt for checkout.
Google 已开始提供 AI 摘要、文本生成可视化报告、同类基准比较与 Ask Advisor，帮助营销人员从数据转向行动。
Google is adding AI summaries, prompt-built visual reports, peer benchmarks, and Ask Advisor to help marketers move from data to action.
该功能已面向美国自助式 Amazon DSP 广告主的认知型活动开放，可生成跨流媒体电视、视频、展示和音频的建议及触达预测。
The feature is available to U.S. self-service Amazon DSP advertisers for awareness campaigns, with recommendations and reach projections across four formats.
Dreamina Seedance 2.5 已开始向部分市场的部分付费广告主推出，支持最多 50 个多模态参考和按时间点控制画面。
Dreamina Seedance 2.5 is rolling out to select paid advertisers in select markets, with up to 50 multimodal references and timestamp-level direction.
Meta 正在美国、加拿大和印度向创作者推出个性化助手，用于解释内容表现、寻找趋势和生成下一步创作建议。
Meta is rolling out a personalized creator assistant in the U.S., Canada, and India for performance explanations, trends, and content ideas.
Meta 正把 Business Agent 扩展到全球不同规模的企业，可在 WhatsApp、Messenger 与 Instagram 回答问题、推荐商品、预约和筛选线索。
Meta is expanding Business Agent globally across WhatsApp, Messenger, and Instagram for answers, product recommendations, bookings, and lead qualification.