ChatCut

• 1. overview (566 chars)

  ChatCut is a browser-based AI video editing tool. Its core positioning is to "use conversations to edit" rather than "use text to generate videos." It combines natural language instructions and transcribed text editing into a truly continuously editable multi-track timeline, greatly improving the efficiency of initial editing of long videos such as oral broadcasts, interviews, and Vlogs. The product has accumulated about 80,000 users and a Discord community of 25,000 people. Subscriptions start at US$25/month. The overall evaluation is biased towards "like a human cut" but it still has shortcomings in long video performance and collaborative engineering.

• 2. Product background (698 chars)

  ChatCut was developed by an overseas team and takes the route of "Agent first" rather than "traditional editing software plus AI buttons". The official positioning of the product is for everyone who records self-recorded content in front of the camera (YouTubers, educators, marketing teams). The slogan is "Put the material in, say what you want, and AI will take care of the rest." It can be run purely in a browser, and also provides plug-ins that adapt to ChatGPT and Codex desktop (installed and driven by the programming agent through the local MCP server), which is relatively rare among AI editing tools. The speed of product iteration has been mentioned many times, and some users reported that "the generation speed and effect are much improved compared to a month ago," indicating that the team is still polishing it quickly.

• 3. Product functions and usage experience (1915 chars)

  ChatCut's editor consists of several core panels: Agent dialogue panel (describing editing requirements in natural language), player, material library, multi-track timeline, export and settings menu. Its workflow revolves around "transcription as timeline": After uploading oral broadcast or interview materials, click Transcript to generate a transcript with timestamps that can distinguish the speaker. Each word is aligned to the screen, so if a certain sentence is deleted, the corresponding video and audio will be automatically cut off - this is the so-called "cutting the video like editing a document."   There are several types of abilities based on this. The first is natural language editing. In the Agent panel, write "remove repeated take", "cut the cold scene" and "tighten this section". ChatCut will call the corresponding tool and implement the changes on the real multi-track timeline; you can also use the @ symbol to specify a certain material. The second is preliminary cut assistance, which analyzes the material, identifies highlights, eliminates duplicate shots, and creates a rough cut that can be further refined. The third is subtitles and supporting materials. Subtitles in more than 100 languages ​​are generated from transcription. It supports multiple subtitle styles and can directly change the text. The fourth is AI motion graphics, which can generate editable components such as chapter cards, charts, timelines, accent effects, lower thirds, etc. on the track with just one sentence, without the need for keyframes or plug-ins. The fifth is supplementary materials. You can use Seedance 2.0 to generate B-roll videos up to about 15 seconds within the project, use GPT Image 2 and Nano Banana 2 to generate images up to 4K, and generate royalty-free soundtracks according to duration. Sixth, it is export compatible. The finished product can be exported to video files or XML and entered into non-linear editors such as Premiere and DaVinci Resolve for further refinement; the Pro solution can also export dynamic graphics in ProRes 4444 with real transparency channels.   In a word, the experience is: what it provides is not "automatic production of a fixed template", but a real timeline that you can take over and modify at any time.

• 4. Pricing and business model (894 chars)

  ChatCut adopts free starting plus points (credit) subscription. The free version provides 25 starting points, no card binding, basic editing and exporting, and the ability to try out AI images, subtitles, music, and motion graphics. The public price of the paid version: 100 points is US$25/month, 400 points is US$100/month, and annual payment is claimed to be 50% off. The consumption items of points include AI Agent rounds, video and image generation, motion graphics, soundtrack, dubbing and rendering; while uploading materials, automatic transcription, manual timeline editing, preview playback, and subtitles generated from existing transcriptions do not consume points. In other words, it costs almost nothing to rely solely on "deleting subtitles and correcting oral broadcasts". Once a large number of generation abilities are used, the points will be consumed very quickly. The official blog also demonstrates templated workflows for marketing teams such as App promotional videos, URL-to-advertising videos, and explanation videos.

• 5. User reviews (1499 chars)

  Positive feedback focused on "saving time" and "looking like a human being." Many users said that they were very satisfied with the 4 and a half minute oral broadcast cut to 30 seconds for the first time and paid immediately. Some people used it to make product PVs, boasting that the generation speed and positioning accuracy (accurate to a few seconds) were significantly improved compared to a month ago, and it also supports BGM generation. There is also an editing novice who tried N tools in one night and strongly recommended ChatCut. He thought its "long to short" recognition is very good and the path is clear and not fancy. Plug-ins adapted to WorkBuddy, Trae, and Codex also surprise some users, who feel that they can cut directly in the familiar Agent environment.   Negative feedback is also very specific. Some users pointed out that the upload speed is average, and it takes more than ten or twenty seconds to upload material of more than ten seconds; the accuracy of subtitle recognition is not outstanding compared with similar products; currently it is mainly good at reorganizing or streamlining long videos, and the joints are basically simple splicing, and lacks the splicing animation and background sound options like editing; they also hope to support the saving of historical versions and clarify whether membership is required for export. An even more pointed one comes from a user who did the couple pause challenge in "Zootopia 2": Agent understood the need, but the built-in tools were limited and could not restore the effect according to the reference picture. In the end, the color draft was published and he was asked to edit it himself. The experience gap was obvious, and the agent complained that the upper limit of a single upload of reference pictures was only 5.

• 6. Industry evaluation (921 chars)

  In horizontal evaluations, ChatCut is usually included in the "2026 AI Video Editor" together with Descript, Opus Clip, CapCut, Runway, VEED, etc. Its differences have been repeatedly pointed out: Descript is a "document-driven plus cloud-based transcription" podcast post-production tool, Opus Clip is a single-point tool for "cutting long videos into short videos", CapCut is a "multi-track timeline plus template ecology" and the strongest mobile terminal, while ChatCut is the representative of "conversational Agent as the main interaction". In a blind test of multiple tools using the same 30-minute interview question, ChatCut was regarded as one of the closest "human editing" effects due to its "solid understanding and film quality, and ability to smooth out single person/question and answer rhythms." However, it is prone to failure in event detection, strict duration or number of segments constraints, and the integration and editing of multi-source content. The overall reputation is that it is "easy to use but has limitations" rather than a "universal substitute".

• 7. Controversies and Risks (836 chars)

  Several risk points have been mentioned repeatedly. First, the point pricing is easy to calculate: 166 seconds of Seedance video, 454 GPT images, and 500 dynamic graphics are all squeezed into the same 100 point pool. Heavy users will quickly use up the Standard file. Second, pure browser (or plug-in) operation is obviously slower than desktop non-linear editors such as Premiere and Resolve when facing large 4K projects and multi-hour materials. Third, the product is relatively new, has a short history, and has no offline mode. The "audit trail" of the decisions made by the Agent in an editing is also limited, and is not as traceable as manual editing. Fourth, some capabilities are still limited by version and built-in tool boundaries, and it is recommended that complex visual restoration needs be transferred back to professional software. None of this constitutes a flaw, but it does mean that it is more suitable as a "first-cut accelerator" rather than as a "full-process end point."

• 8. Applicable groups and usage suggestions (799 chars)

  ChatCut is most suitable for three types of people: YouTubers and educators who record themselves in front of the camera, marketing and product teams who need to quickly produce interviews/spoken broadcasts, and editing novices who don’t want to learn Premiere but want to make decent post-production work. It is especially good at "making long videos shorter" and "reorganizing clips semantically". Not suitable for: short video special effects players who pursue cool transitions and fine splicing effects, teams that require stable offline work or strong collaborative version management, and highly constrained tasks that require pixel-level restoration of reference images (such recommendations go directly to professional NLE). As for alternatives, choose Opus Clip for pure short-cut videos, Descript for podcast post-production, CapCut for lightweight creation on mobile devices, and Runway for generative special effects.

• 9. Summarize (270 chars)

  ChatCut has made "conversational agent plus real editable timeline" its own moat. It is a great tool for saving a lot of time in the initial editing of long videos. As long as you use it as an accelerator rather than an end point, and try out a representative project to clear your points account first, it will be worth the money.

No in-depth report yet

• 

  EugeneWardK

  —

  它真正值钱的地方是把「对话式 Agent 加可编辑时间轴」做成了护城河。我用自然语言说「去掉冷场、加字幕、导出 9:16」，一条 prompt 就跑了转写、找高光、剪时间轴、压字幕全流程，导出还能走 XML 进 Premiere 接着精修，对长视频初剪是真能省大把时间的利器。

---

• 

  Judy167

  —

  回不去了，剪视频就像聊天一样。

---

• 

  Scott.Fisher_Max36

  —

  昨天拿 ChatCut 做产品 PV，生成的速度和效果都比一个月前提升非常多了，而且支持生成 BGM，也能精确定位到第几秒。之前用剪映手工拖时间轴，一个 3 分钟的 PV 要折腾大半天，换成 ChatCut 后丢进去说一句「做个产品宣传片」，agent 自己找高光、加字幕、配乐一条龙，我只在末尾微调了几处字幕识别的小错就导出了。对我这种不想学 Premiere 的人来说，对话式剪辑真的省了太多时间。

---

• 

  purplepanda291

  —

  积分池这个设计真让人头大，生成视频和动效抢同一百个积分，重度的很快用光。我一个月前还在 Standard 档悠哉悠哉，这月开始集中做系列短视频，又是 Seedance 生成 B-roll、又是 Nano Banana 出图、还要叠动态图形，100 积分一周就见底，最后只能临时升 Pro，账算下来比预想贵出一截。

---

• 

  墨染476_1

  —

  我装了 ChatCut 的 WorkBuddy 插件，拿一段 7 分钟的口播教程试剪，整体体验相当强。它给了一个在线预览的剪辑界面，不满意的地方能直接人工改完就导出。唯一问题是英文发音不标准时字幕识别会小翻车，但其他已经很强大了，真心推荐试试。

---

• 

  Billy.Peterson_Plus85

  —

  第一次把 4 分半的口播剪到 30 秒，太香了。

---

• 

  RGonzalezSr1

  —

  单 25 美金的档位把剪辑、字幕、配乐、动效都包了，差不多顶掉我之前三四个订阅，算下来还真划算。

---

• 

  LaurenWatson_2020165

  —

  拖 transcript 里的字条就能改视频前后顺序，这个设计太顺手了，找经典片段集锦特别快。

---

• 

  Maddox19

  —

  参考图让生成的镜头和真实素材一致，这个细节做得到位。

---

• 

  EFosterII

  —

  做《动物城 2》那种参考图还原的需求它目前还搞不定，agent 理解了但内置工具受限，最后还是得回剪映，希望新版本能把内部工具撑起来。

---

• 

  PenguinPoolGonzalez

  —

  界面是真的友好，小白也能直接上手。

---

• 

  安然297_1

  —

  不能保存历史版本，这点真的恼火。

---

• 

  Gary.Howard_Pro312

  —

  横向实测下来 ChatCut 确实最像「人剪」，能理顺单人问答和 Vlog 的节奏，结构干净自然。但严格时长或段数约束的任务它容易失手，长视频和批量也略吃力。当初剪加速器用很香，别指望它一步到位当全流程终点。

---

• 

  RSmith_Pro32

  —

  字幕识别没想象中准，跟剪映比没啥突出优势。

---

• 

  Sandra40

  —

  纯浏览器跑 4K 大工程还是明显比桌面 NLE 慢，长素材建议先粗剪再导出去精修。

---

• 

  HalvingHawkMitchell

  —

  导出要会员？这个卡点设计的也太巧妙了。

---

• 

  MsYouandiHommersom_pro

  —

  衔接处目前基本是简单拼合，想要剪映那种炫酷转场和拼接动效还得再等等，希望后面版本补上。

---

• 

  毛素_1

  —

  上传速度一般，我传十几秒的素材也要等十几二十秒，希望能再快一点。

No reviews yet