用验证码登录,不必记密码
登录只剩一个输入框。填入邮箱,我们发一个验证码让你进来;习惯用密码的仍然可以用。Google 和 Facebook 按钮已经移除,进门路上不再经过别家公司的页面;如果你之前用 Google 登录,同一个邮箱仍然进到同一个账号,做过的东西都还在原处。
Brainpercent 的每一次变化,作者 Edward Ilin。最新在前。
登录只剩一个输入框。填入邮箱,我们发一个验证码让你进来;习惯用密码的仍然可以用。Google 和 Facebook 按钮已经移除,进门路上不再经过别家公司的页面;如果你之前用 Google 登录,同一个邮箱仍然进到同一个账号,做过的东西都还在原处。
现在有了联系页面,消息直接送到 Edward,而不是排进工单队列。你可以选择这封信是关于什么的,包括安全反馈或故障,这样任何一类问题都不需要单独的邮箱才能找到对的人。安全页面也指向这里。
博客有了真正的首页:一篇头条、下面一片网格,以及按分类的筛选,分类是从真实存在的文章里生成的,所以筛选永远不会给出一个空页面。没有配图的文章现在会用上站内同样的水洗封面,页面底部还有一个投稿的入口。
另有 8 项小修复。
现在定价页面和应用内共用同一份价目表,而它给出的信息远不止一个数字。每一项都会告诉你能得到什么、怎么开口要,以及要花多少,每一项还配有示例,你可以直接复制到聊天里试试看。免费的操作会标注为免费,只有在成功生成结果时才会向你收费。
每张套餐卡片现在都会显示未使用余额有多少会结转到下个月:Starter 10%,Plus 20%,Pro 30%,Business 50%。积分现在在整个产品里也有了自己的专属标志,而不再用美元符号,这样余额读起来就像是你还能做多少工作,而不是账户里在流失的钱。你直接购买的积分永远属于你,且永不过期。
当你请求社交帖子时,一张实时卡片现在会展示每条帖子的生成过程,让你随时知道进展。生成过程中你可以继续聊天,帖子完成后随时打开成品或前往素材库查看。通知现在也会直接带你打开对应的帖子。
写在每张帖子图片上的标题现在会遵循你的 Brand Author 声音和你的业务语言,而不是默认使用通用英语,因此希伯来语品牌会得到一个希伯来语、符合品牌调性的标题。同一个修复也让帖子正文本身恢复了品牌声音,此前它一直在悄悄地脱离你的设置运行。
希伯来语现在使用专为其打造的字体显示,积分花费会以简洁的行内形式呈现,例如(1 积分),回复中多余的破折号也已经消失。
删除项目现在会先弹出明确的确认提示,然后移除与该项目相关的一切,包括内容、排期和设置,不会留下半删除的残留状态。
现在创建项目时,每个排期和定时器都默认关闭,除非你自己手动开启,否则不会自动发布或运行任何内容。
在聊天中提出视频请求,就能得到一条从头到尾规划并生成的完整多场景视频,而不只是单个镜头。
无需离开聊天窗口,就能把任何想法或文档变成一张干净、忠于原文的信息图,可选六种风格。
现在无论在聊天还是在素材库中,你都可以为视频叠加音乐、配音和字幕。
拖入一个文件,立即在聊天中用它制作视频、编辑图片或合并素材,无需单独的上传步骤。
一个简单的切换器让你选择撰写内容的声音,这个选择现在会贯穿你生成的一切,从帖子、回复到文章。
社交编辑器现在会显示帖子的发布状态,可以链接到已发布的帖子,并能一键取消发布,所有操作都在同一个地方完成。
现在你可以把团队成员添加到指定项目中,并且只能通过邮件把内容发送给已验证的地址。
登录前打开应用的用户,现在看到的不再是一片空白,而是一个清晰的引导式空状态,以及一个展示可用功能的锁定工具栏。
我们的对比页面和使用场景页面现在已在全部 13 种受支持语言中撰写完成并可被搜索引擎抓取,让更多人能用自己的语言找到 Brainpercent。
素材库和评论现在采用了和关注列表一样干净、适配手机的布局,可以快速筛选出真正值得你关注的内容。
年龄政策现在是一条清晰的规则:18 岁及以上,在服务条款、隐私政策以及 13 种语言的每个注册勾选框中表述一致。同一次诚实审查还删除了我们无法兑现的承诺,包括一项客服响应时间承诺,并修复了两个未能记录用户同意的注册路径。
Cookie 政策、可接受使用政策、安全概览、无障碍声明、内容举报、公开的法律变更日志,以及全面重写的退款政策。子处理商名单从 14 家扩展到 27 家具名供应商,每个页面都以全部 13 种受支持的语言呈现。
十一个在手机上原本没有入口的页面现在都有了导航路径。聊天打开时是一个干净、聚焦的主屏,控件在窄屏上不再被裁切,整个应用的触控目标都为手指优化。
语言纯度关卡现在会拦截混合语言的输出,希伯来语帖子再也不会夹杂英文连接词。写作引擎遵循每种语言的母语文案技艺,你的品牌记忆和业务目标同时驱动初稿和每一次改写。
在聊天中要一条视频,它会把五个相互衔接的镜头规划成一次连续生成,没有拼接痕迹。你先审阅方案,批准后才扣除积分,成片一步即可变成社交帖子。
聊天中生成的图片会进入你的素材库,并可一步变成帖子。每个条目都显示其工作流状态,内容按项目严格隔离,删除就是真的删除。
浏览创作者和带有真实互动数据的爆款短视频,标记适合你品牌的内容,工作室就会用你的产品和信息重现该形式,并在构建过程中实时显示进度。
聊天界面、编辑器、通知和导航现在能在从右到左的语言中正确镜像,时间戳在双向文本中依然清晰可读。RTL 内建于布局之中,而不是在翻译之上刷一层。
新的发布身份选择器让每条帖子既可以以你个人的名义发布,也可以以公司主页的名义发布,使用你已连接的组织账号。
声量份额追踪器对比你和竞争对手被引用的频率,新的 LLM 提及集成则基于真实查询数据,找出语言模型在哪些地方提到了你的品牌。
以键盘为先的命令面板现已覆盖整个产品,分步引导教程带新用户熟悉应用。两者均已完整翻译。
技能现在通过一次明确的选择来激活,不会有任何东西被悄悄添加,选择器保持紧凑而不会溢出输入框。技能目录向所有用户开放,没有即将推出的锁。
我们彻底移除了 OpenAI 集成。Brainpercent 撰写的每一段文本都由 Anthropic 的 Claude 模型生成,图片、音乐和视频则通过专用媒体网关处理。一套文本技术栈,一种一致的声音。
在聊天中要一段音乐或配音,即可在对话里直接播放结果。带人声的歌曲、换脸和音频转录也在同一周加入了创意工具集。
另有 10 项小修复。
A new War Room tab continuously ranks the posts worth responding to right now, drafts an on-message reply grounded in your positions, and shows it with a verification report (facts, on-message, risk) so you can approve, edit, or reject from your phone. Nothing publishes without your tap.
We pull the actual content of each post (and its images, with click-to-preview) instead of a thin search summary, so you can read the full context. When you generate a reply, the assistant now reads the post's image too, so the draft is grounded in what the post actually shows.
Create your own skill-agents, or install ready-made ones from the catalog (including a new campaign pack), and they connect straight into your chat as slash commands. Skills automatically ground in your brand's positions and voice, and your new-chat page now shows the skills you have connected.
War Room, Analytics, and the daily plan now share the same clean starting screen with a single button that runs the full AI setup for that surface, so an empty page becomes a working one in one tap instead of a dead end.
On longer fields in the ASO setup and the brand-template editor, every keystroke was triggering a re-render of the entire surrounding panel. That was fine on a fast laptop, but on an iPad you could feel keys trail behind your typing. We now buffer keystrokes locally and only push the value up after a 200ms pause or when you leave the field, the same fix already in place for the social editor caption. Typing is instant on iPad again.
When you generate or regenerate an image, video, or apply branding inside the social editor, the loading overlay used to look different on every tab (Instagram, LinkedIn, etc.) and the labels on Instagram were always in English even if you used the site in Hebrew or another language. Everything is unified now: same spinner, same wording, and all labels translate into your selected language across every tab.
When you Generate Video on a post, the clip used to come out with audio that just read the post title out loud, which was generic and off-putting. Now we send the post's topic and platform to an AI sound director that returns a one-line music brief (genre, instruments, mood, tempo) tailored to the content, and the video engine builds purely instrumental audio against that brief. A snappy product post sounds punchy; a calm guide sounds reflective; nothing is ever narrated.
Our sitemap tells search engines which pages on the site are worth indexing. It now strictly lists only articles from Brainpercent's own blog projects, and skips anything generated for customer accounts. One customer article had been silently leaking through; that's gone now.
Cleaned up where every language URL goes. The blog pages in each language used to redirect back to the English blog by mistake; they now serve their own language directly. The newsletter and a few other English-only pages now send you somewhere real from any language, instead of a missing page. The sign-in flow no longer sends a conflicting signal to search engines. Together this clears around three quarters of the indexing warnings we had pending.
If a Spanish article is shared somewhere with the English URL, the site now sends the reader straight to the Spanish version of the article instead of serving the same article at both URLs. Same for every supported language. Google was treating these as duplicate pages, which hurt how often any version showed up in search; now there is exactly one canonical URL per article.
The site map sent to Google was missing the features page, every language version of the features page, and the Hebrew about, contact, and white-label pages. Newsletter issue pages were also pointing at a slightly different domain than every other page on the site, which made Google treat them as duplicates. Both are fixed. We also cleaned up an unused Ukrainian language code that was sitting in the routing config without a real page behind it. None of this changes anything you see, but it should improve how the site is discovered in search.
When you draft a reply to a post in Opportunities, it now argues from your business's goals and positions. If the post is from a competitor or pushes a view you oppose, the reply counters it from your stance instead of politely agreeing. On X, a reply from your own account now speaks in your voice and takes a clear position, while still keeping it sharp rather than abusive.
Trending picks are now judged against your real agenda and your audience, not just whether a headline mentions your keywords. A story that name-drops your topic but is really about something off-mission (a markets or lifestyle outlet covering your subject for a different angle) is dropped, even if it is a huge story. You will see fewer picks sometimes, but every one is something you can post about authentically. It is better to surface three stories that fit than ten that don't.
When you generate from a story, the writer no longer just summarizes it in your voice. It reacts to it from your goals and your positions, using your business memory as the source of truth. If the source pushes a rival or argues the opposite of what you stand for, your post rebuts it instead of repeating it. A final safety check catches anything that slipped through and rewrites it from your side. The result: every post works to grow your business, never the competition's.
In Opportunities you can now open any conversation and read the full post, who wrote it, where, and when, without leaving Brainpercent. A Show more button loads older items, the day filters (Today, Yesterday, This week) line up with the real dates, and you can now follow X sources alongside the rest.
On X you can now publish a real multi-post thread, not just a single tweet. When you generate from a trending story you get a Single or Thread choice, and every post in the editor has the same switch, so you can flip between a single post and a full thread anytime. Thread mode drafts a hook-first thread grounded in the entire source article: one idea per post, each post opening a loop into the next, and a closing post that invites a reply instead of ending on a plug. It publishes as a native X thread through your connected account, and if you set a first comment it rides along as the final reply.
Generate Video in the post editor now animates the clean image, never the branded one, so your headline text no longer bleeds into the video. Every clip also comes with native AI audio (ambient sound and music) so it doesn't feel flat. Branding a video is still a separate one-click step if you want it.
When you generate a post or carousel, the image now arrives clean, with no branding baked on. If you want your logo, colors, and headline on it, click Brand (or Brand All for a whole carousel) and it stitches in one step. This applies across the editor and the posts created from Trending, PR, and your Daily tasks. Before, carousels were branded automatically the moment they finished, so there was no clean version to work from.
Click an image in your Media library (or say "animate this" in chat) and it becomes a short video. Three upgrades landed. Animating a single image is much faster than before. It animates the exact image you picked instead of guessing at a different one. And it actually looks at the picture first, so the motion matches what is in it (an ocean shot gets rolling waves and a slow push-in, a portrait gets gentle drift) rather than inventing an unrelated scene. The video also renders at the image's own shape, so a square or landscape image is no longer forced into a vertical crop. It animates the clean image so any branding text stays crisp.
Both tabs now have a single Set up button that configures everything from your project's business profile, with no typing and no picking from a generic list. On Trending, one click follows the news feeds that match your niche and audience. On Opportunities, one click finds the communities where your audience asks questions and sets them up for you. The empty state was cleaned up too: one clear primary action instead of a wall of buttons and irrelevant suggestions.
The Analysis tab (right below Daily tasks) is rebuilt to match the Daily tasks walkthrough, but for research. One "Run full analysis" button runs five same-day reports in sequence, each also runnable on its own: a Brand intelligence report (positioning, audience, strengths, opportunities, risks), Competitor SEO keywords with the gap angle to win each, GEO keyword research (the real questions people ask AI in your niche and how to become the cited answer), an AI visibility check (how your brand shows up when people ask AI about your category today, and what to fix), and Competitor review pain points with the opportunity to win on each. Every report is generated in your project's language and grounded in your business profile. A calendar-style day navigator steps back through saved days, and Export PDF turns the whole day's analysis into a clean printable report. Each report self-heals with an automatic retry so a transient hiccup never leaves you with a blank.
The "Daily tasks" surface (renamed from Today, now right below Search) is a step-by-step console. One "Run full plan" button runs every task in sequence and fills the page: Social pack (a multi-platform pack from today's top topic, with selectable channels), PR & press release (from the latest verified news), Trending intel, Blog article (suggests daily SEO keywords for your business, one click to generate an article), Comment drafts (up to 20 ready-to-paste reply drafts), and Outreach (first-touch emails for prospects you haven't contacted). Every task is an on-demand generator with its action button beside a Mark-done toggle, a timeline you walk top to bottom, and a calendar-style day navigator to step back through saved days. Also fixed: the morning run now seeds the social pack from your RSS feeds when the trending cache is cold (it used to silently skip everything), reply drafts now pass an achievable quality bar, and the social pack generates in the active project's language.
Apply in seconds from /chat (an AI classifier auto-approves anyone with a credible audience). Get a tracked link, a one-pager Marketing Kit, and a live dashboard showing clicks, signups, and commissions. Payouts settle on the 1st of each month via Stripe Connect. Once you cross $50 we wire it straight to your bank. Includes self-referral detection (Stripe card fingerprint), 7-day cooling-off on new accounts, and a 3-day refund hold so clawbacks are clean.
Privacy and Terms were rewritten to cover the EU AI Act, the Digital Services Act, sanctions rules, disclosure about what AI can get wrong, a 14-day cooling-off period, and what stays true after you leave. A data processing agreement is available pre-signed for companies that need one. The cookie banner now asks before anything measures you, and you can export everything we hold about you from your settings: it arrives by email and is deleted again after ninety days.
Cold outreach moved into the chat surface end-to-end. Prospect/research/draft cards inline, edit mode in-place, auto-unlock emails when research finishes, batch-sequence prospects with one command. The old admin form is retired.
Every kind of generation now costs the same: one credit for an article, a social post, an image, a video. Carousels are the one exception, because a carousel genuinely makes several pieces. If a balance will not cover what you asked for, you are told before anything starts rather than after, and topping up takes a moment.
Boost any post to Pinterest with a real board picker (no more guessing the destination). Zernio publishing now aligned to all 13 supported channels (Telegram + Snapchat added). Caption length safety net for every platform so a Twitter clip never silently truncates a LinkedIn post.
Image prompts now carry a per-channel aesthetic suffix. Instagram gets editorial-magazine, LinkedIn gets corporate-clean, TikTok gets vertical-energy, X gets newsroom-sharp. Headlines also persist per-platform now, so editing the Instagram title doesn't accidentally overwrite Twitter.
First pass made every carousel slide a centered hero magazine cover. It looked impressive and it killed the storytelling rhythm. Slides now share a cohesive DSLR / photoreal / Wealth-quality DNA but compositions vary by role: hook = movie-poster hero, insight = close-up detail, story = wide establishing, breathing = atmospheric still life, outro = reflective close. WIDE → CLOSE → CLOSE → WIDE → CLOSE rhythm enforced across every plan.
The image-gen prompts told the model brand colors MUST be the dominant tones throughout, with nothing holding them to reality, so images came back as glowing purple particles. Hardened the wrapper across 5 generation paths (manual, autopilot, regenerate, blog hero, IG carousel) with HERO COMPOSITION, hard warm rim light, crushed blacks, Wealth / Time POY / Hodinkee / Bloomberg reference, and a hard ban on cyberpunk neon. Brand colors now appear as a single subtle accent, never the entire mood.
Single URL to share on launch day. 24 SaaS directories listed with one-click upvote/review buttons, status badges (LIVE / MAY 4 / PENDING), and a direct link to /pitch/demo. Editorial design matching the pitch page.