Picture a countdown: if your TraeWork client is still on the old version, twelve days remain. After October 23, unupgraded clients lose access -- a deadline stated in ByteDance's official documentation, as reported by pandaily. Behind the countdown sits a bigger move announced on October 9: TraeWork and TraeCode are officially merging into a single new TRAE, upgraded into a full-chain development platform. In one sentence: two tools that each handled a slice of work -- one for AI office tasks, one for AI coding -- now live inside the same TRAE.
This is more than product-line housekeeping. Unpack the official formula of "one entrance, two modes, three synced devices" and you find a bet about how work is changing: the boundary between office tasks and coding is dissolving, and a task should flow from idea to deliverable without switching products. This article lays out the official account, the migration checklist, genuine community feedback -- including the complaints -- and the competitive context.
One Entrance, Two Modes, Three Devices: The Official Breakdown
The announcement describes the new TRAE along four lines, and each deserves a close look.
One entrance: requirement breakdown, solution drafting, coding, test runs, documentation, and deliverable production all happen inside the same TRAE. Previously you might generate a proposal in TraeWork, then switch to TraeCode to turn it into code, moving work across by hand. Now both halves share one building, and deliverables never change address from birth to final version.
Agent mode: from idea decomposition to delivered results, with support for dispatching multiple agents in parallel across projects. One agent can rebuild a page in project A while another compiles a test summary for project B, without stepping on each other. This mode targets a "describe the outcome, get the result" workflow, leaving the process details to the agent.
IDE mode: the full intelligent development environment remains, aimed at deep single-project coding, debugging, and fine-grained edits. This is the tier for people who need to watch every line -- refactoring a core module or chasing a stubborn defect is not work you hand off entirely.
Three synced devices: desktop, web, and mobile clients stay in step. The official example runs like this: start a task on your computer, check progress and add requirements from your phone during the commute, then pick up where you left off back at the desk. Tasks follow the account, not the device -- a practical change for anyone who works across locations.
Assembled together, these form the skeleton of the new TRAE: the unified entrance guarantees everything lives in one place; Agent mode lets work run on multiple tracks; IDE mode preserves control down to the smallest detail; and device syncing means the work follows the person.
Shutdown Countdown and Data Migration: What to Do Before October 23
For existing users, the timeline matters more than the product philosophy.
Per the official announcement, TraeWork shuts down on October 23, 2026 -- the date comes from ByteDance's official documentation as reported by pandaily -- and users who have not upgraded will lose access. TraeCode users receive the update automatically; TraeWork users must upgrade their clients manually. Rollout proceeds in gray-release batches, and the official notice says explicitly that not receiving the update immediately is normal. In other words, your upgrade window is not entirely your own: if the batch has not reached you, waiting is the only option, and if it has, delay risks colliding with the shutdown line.
On data, the official migration checklist covers chat history, account data, profile information, login state, local data, and business credits -- all fully preserved and migrated automatically. Nothing needs exporting or importing; the single critical action is completing the TraeWork client upgrade before October 23. Do not leave it to the final day: gray-release timing plus a hard shutdown date make a one-day buffer the minimum sensible insurance.
The merged TRAE also carries forward a set of capabilities: automation tasks covering weekly reports, routine checks, test summaries, and report generation; a template library that turns common workflows into starting points; a personal assistant for day-to-day errands; and a project deliverables hub where code changes, HTML, PPT, Word documents, images, video, and reports can be viewed and iterated in one place. For users who previously hauled outputs between two tools, this may be the most noticeable improvement of the whole merger -- deliverables finally have a single home.
Why Merge Office and Coding Into One Platform
The real question behind the merger is why ByteDance restructured the product line instead of letting the two tools coexist.
The first reason lies in how work itself has changed. Over the past year the border between AI office work and AI coding has blurred: writing a weekly report is office work, compiling a test summary is office work, yet the test results come from coding tasks. Drafting a product proposal is office work; turning that proposal into a running page is coding. In real projects the two are upstream and downstream on the same assembly line. Once AI can carry the whole line, selling tickets to its two halves separately stops making sense.
The second reason is competitive positioning. Our August 2026 survey of the AI coding agent landscape (coding-agent-landscape-hotspot) noted that Trae then competed in the local IDE camp alongside other tools. Across the industry, every player is expanding coverage: IDEs reaching for the terminal, terminal agents reaching for deployment, office suites absorbing documents and reporting. A full-chain platform means a user enters once and travels from idea to delivered product; every additional stage covered raises the cost of leaving. Merging the two products folds two markets' tickets into one, setting TRAE directly against rivals chasing the same full-chain vision.
The third reason is context. Requirement discussions and code lived in different tools, so user context was split: the proposal document on one side, the repository on the other, with AI re-learning both ends each time. After the merger, one conversation can carry from requirement description to code implementation within a single context -- an invisible but real gain for generation quality.
There is also a quieter implication for people who never write code at all. The automation tasks bundled into the merged product -- weekly reports, routine checks, test summaries, report generation -- are office work by anyone's definition, yet several of them feed on coding output. A product that only handles documents has to stop at the boundary; a product that owns the code side can follow the thread across it. ByteDance is effectively betting that the next wave of office AI adoption will be pulled by deliverables that include software, not just slides and spreadsheets. If that bet lands, the merged TRAE serves two audiences with one product: developers who want the IDE depth, and operators who want outcomes without ever opening a code editor. If it misses, neither audience gets a tool tuned tightly to its needs, which is precisely the risk every "everything platform" carries.
SOLO, Cue, Rules, and MCP: The Coding Foundation Stays
Developers worried that the merger might dilute coding capability can relax; the official account keeps the coding arsenal intact.
SOLO mode ships with a dedicated coding agent that understands goals, plans tasks, and orchestrates tools, advancing each development stage independently -- closer to an engineer who runs the whole relay than an autocomplete tool. Cue, the completion layer, was upgraded alongside: it senses repository context and editing trajectory, predicts the next change point, and applies it with a Tab keystroke, supporting multi-line edits with diff previews. One layer covers "hands off and let the AI work"; the other covers "I am driving -- keep up." Together they serve both postures.
For customization, TRAE Rules lets users configure their own AI working rules -- team conventions, code style, and document formats can all be written down and enforced; MCP handles external tool assembly, wiring third-party capabilities into the workflow. Teams with strict toolchain requirements should examine these two features first when estimating adaptation cost. Within the evaluation framework of our IDE comparison (ai-coding-ide-comparison-review), both count as differentiators. A follow-up hands-on piece on Agent mode is planned, covering how to put the dual modes and these configurations to real use.
Community Feedback: The Default Agent Workspace Isn't for Everyone
The merger has not drawn applause only. Genuine feedback from the official community announcement deserves an honest airing:
- After upgrading, the app opens the Agent workspace by default; switching to IDE mode requires a manual toggle, and no option to change the default startup mode currently exists.
- Some users report that guidance for switching between the two modes is not obvious, leaving new users unsure which side they should work in.
Both complaints point to the same product judgment: ByteDance treats Agent mode as the flagship post-merger experience, and the default is the announcement. But for coding-heavy users, manually switching to the IDE every morning is a friction that compounds. A simple "remember my preference" option is not hard to build; whether it arrives is worth watching -- it signals whether ByteDance sees IDE users as core assets or migration baggage.
Viewed over a longer horizon, this controversy is a predictable stage of any tool merger: incumbents demand continuity while the new product demands a unified narrative. The good news is that both modes survive and IDE capabilities were not cut; the dispute sits at the level of defaults and onboarding, which fast iteration can resolve.
The Competitive Board: Full-Chain Platforms Are the Next Battleground
Placed on the wider table, the full-chain development platform is becoming the direction major players bet on together. Our analysis of the meta war among coding agents (meta-coding-agent-war-hotspot) mapped how differently the big names push: some press down from the model layer, some climb up from the terminal, some entrench in the IDE. ByteDance's play is addition of a different kind -- rather than launching yet another product, it welds two user bases and their scenarios together under one name, claiming the whole "office plus coding" territory.
The three-device sync also holds a lesson for the terminal-agent camp. Terminal agents excel at autonomous execution but struggle with visibility while the human is away, a point our terminal coding agent comparison (terminal-coding-agent-comparison-review) examined: what you can see once a task is running, and how easily you can intervene, decides whether you dare hand it anything important. Checking progress from a phone and adding requirements on the move is a direct answer to that problem. For anyone still choosing tools, our 2026 scene review (ai-coding-tools-2026-scene-review) and AI coding agent comparison (ai-coding-agent-comparison-review) are worth reading side by side -- under the merger wave, "which product" is turning into "which ecosystem."
Timing deserves a note of its own. Setting a hard shutdown date barely two weeks after the merger announcement is an aggressive adoption tactic: it guarantees a fast migration spike, but it also guarantees that every rough edge in the transition gets tested under pressure. Vendors usually spread such windows across a quarter. ByteDance compressed it, which signals confidence in the migration pipeline -- and, just as plausibly, a desire to close the two-product narrative before competitors frame it for them. Either way, the coming weeks will produce a stream of real-world migration stories, and how loudly the friction gets reported will shape how the merger is remembered more than any launch material will.
Closing
The TraeWork and TraeCode merger looks like product-line tidying on the surface and a wager underneath: ByteDance bets that office and coding will converge fast enough to justify pressing existing users with a hard shutdown date. Before October 23, TraeWork users need one manual upgrade; after that, two products' stories become the story of one TRAE.
Whether the merger stands will not be decided by launch rhetoric but by three things: whether migration proves as painless as the official checklist promises, whether the default-Agent-workspace complaints get answered, and whether the full-chain experience genuinely beats stitching two tools together. Answers will arrive over the coming months. If you are a TraeWork user, put the upgrade on the calendar now; if you are watching from outside, waiting for the gray release to stabilize and community sentiment to settle is a reasonable stance.
One practical caution before committing a team: treat the upgrade as a migration event, not a background update. Verify that your automated tasks and team workflows run correctly inside the new TRAE before the October 23 line, keep the migration window open until it finishes, and keep a local copy of anything business-critical until the dust settles. The official checklist says data moves automatically and completely, and there is no reason to doubt it -- but migration weeks are also the weeks when support queues are longest. Budgeting your own buffer is cheaper than disputing outcomes later.