Release Notes

Archive a Knowledge node together with its subtree, then restore it whenever you need it again.

• Archive Knowledge nodes by subtree to move them out of lists and search results.
• Turn on “Include archived” in the filters to review archived Knowledge and restore it to its original location.

Trusted work can now run without step-by-step approval, and every approval is recorded in an audit log.

• New Auto (no approval) mode and per-thread execution permissions let trusted work proceed without approving each step.
• A permanent audit log records who approved which operation and when, so the history can be reviewed later.
• Browser notifications now show the task name and let you approve directly from the notification.

Claude Opus 5 is now available as a model option in Worker settings.

• Claude Opus 5 is now a selectable model in Worker settings.
• AI Worker evaluations can now run with files attached.

Drag Knowledge images into multiple columns or between paragraphs, and profile editing is easier to fill in.

• Drag Knowledge images into multi-column layouts or move them between paragraphs.
• The profile form in Settings now uses consistent width and descriptions for an easier layout to fill in.
• Approve AI Worker execution from external chat such as Slack or through OAuth (MCP) connections.

Group members into departments and share AI Workers and secrets with a whole department at once.

• Group organization members into departments and configure permissions and resource sharing per department.
• Share resources such as AI Workers and secrets with an entire department from the sharing dialog.
• Deleted AI Workers stay restorable for 30 days, so an accidental deletion can be undone.

Start from templates that handle inbound inquiries (CS Inbox) or monthly accounting work.

• A customer inquiry (CS Inbox) AI Worker template covers responses to public embed inquiries and handoff to a person.
• An accounting AI Worker template takes on monthly closing and journal entry work.
• Workers can safely fetch Cloud Box receipt files and attachments from connected services into their workspace.

You can now choose which optional permissions to grant when connecting an external app.

• Choose which optional permissions to grant directly on the OAuth consent screen.
• Public API scopes and target resources are separated so external apps receive only the access they need.

Knowledge results now show the creator, update time, and folder, while OAuth consent and connected-app management add safer external connections.

• Knowledge search results now show the creator, last-updated time, and containing folder.
• A new OAuth consent screen lets you review requested access before connecting an external app.
• Settings now lists connected apps and granted permissions and lets you revoke connections you no longer use.

An official slide-creation worker is now available, and the worker template gallery makes it easier to choose and understand each template.

• The official slide-creation worker can generate PNG assets and editable HTML decks together.
• The worker template gallery now uses cards, with required integrations and usage guidance on each detail page.
• zemu/default now runs on Claude Sonnet 5, with more reliable sandbox file persistence and recovery.

Save a web page URL as Knowledge and make its content available to AI workers.

• Save web page URLs in the Knowledge tree and make them available to AI workers.
• URL Knowledge works with shared links and the public API and can be organized alongside existing pages and files.
• The mobile app now shares the web app's API foundation for more reliable authentication and chat.

The new "Grok 4.5" model is now available, and the knowledge review UI has been redesigned with line-level comments.

• The new "Grok 4.5" AI model is now available, delivered through a Zero Data Retention (ZDR) environment.
• Redesigned the knowledge review UI. We clarified the distinction between reviews and comments and added a way to comment on a per-line basis.
• Improved automatic naming of chat threads so they get titles that better reflect the conversation.

The model catalog has been refreshed with GPT-5.6 added, and you can now add and manage sub-workers all from the permissions screen.

• Refreshed the model catalog. We added "GPT-5.6" and tidied up rarely used models so you can pick the right model for the job.
• You can now add and manage sub-workers (connected agents) all at once from the permissions screen, with tighter integration with approval settings.
• When a sandbox is recreated, your workspace is restored automatically, and images embedded in knowledge pages are managed as page-owned assets, so references are less likely to break.

The desktop app has been rebuilt to deliver the same experience as the web version, with a unified login screen.

• Rebuilt the desktop app so it works with the same UI as the web version, delivering a consistent experience across desktop and web.
• Unified the login screen with the web version for a clear, consistent sign-in experience.
• The desktop header now switches smoothly depending on your login state.

A public REST API for operating on knowledge from outside is now available, making it easy to integrate with external systems.

• Added a public REST API for operating on knowledge from outside, making integration with external systems easy.
• Added access via MCP and published OpenAPI/Swagger API documentation.

Slash commands, live Mermaid diagram previews, text color and highlights, and more make the knowledge editor far more expressive and easier to use.

• Added slash-command formatting, live preview for Mermaid diagrams, text color and highlights, image resizing, and drag-and-drop file uploads.
• Type "@" in the chat input to search and select knowledge, then attach its content as context for the conversation.
• Version history now shows the date and editor's name, and each thread remembers the AI model you last used and selects it automatically when you resume.

You can now collapse and expand sections in the dashboard sidebar, folding away sections you don't use for a clearer overview.

• You can now collapse and expand each section in the dashboard sidebar. Folding away sections you don't use gives you a clearer overview.
• Adjusted the height of section headings and the spacing between levels, making navigation easier to read.

AI now automatically summarizes long conversations so you can keep chatting without losing what matters. You can also tidy up manually with /compact.

• As a conversation grows, AI automatically summarizes its contents. Important information is preserved, so long conversations stay comfortable.
• Just type the /compact slash command to manually tidy up the current conversation history at any time.
• Uploading large files to knowledge is now more stable, and you can check upload progress in a list.

A new AI assistant drawer you can open anytime from the right side of the screen. External knowledge edits are now more responsive too.

• Added an AI assistant drawer you can open anytime from the right side of the screen. It automatically understands the page you're on and supports you based on your work.
• Edits made to knowledge from outside the editor are now reflected instantly, making everything feel snappier.
• When a permission error occurs with the Google Drive integration, clear guidance for reconnecting is now shown.

You can now choose Opus and Sonnet 5 with the Claude integration. We also fixed an issue with reactivating subscriptions.

• You can now choose "Opus" and "Sonnet 5" with the Claude integration.
• Fixed an issue where reactivating a subscription did not work correctly.

A new mode lets you share your goal before the AI starts working. You can also copy files from the file menu.

• Added "goal-first mode" — share your goal before the AI starts working. Your intent is understood more accurately, so you get better results with less back-and-forth.
• You can now copy files from the file menu, making it quick to create a new file based on an existing one.

You can now talk to AI Workers directly from Chatwork rooms. Assign an AI Worker to each room for automatic replies.

• Chatwork integration is now available. Connect your account and assign an AI Worker to each room — messages are automatically answered by AI.
• The Kimi K2.6 model now supports sending messages with images.

Replays now show the user's input as a typing preview so the original interaction is easier to follow.

• User input now appears as a typing preview in the input field during replay.
• Messages later in the conversation are previewed before being added to the transcript.
• Text-field focus styling is now subtler, reducing visual noise while keeping focus visible.

Knowledge database tool results can now be reviewed in a dedicated right-hand panel. Signed-in users can also browse the top page freely.

• Results from knowledge database tools now appear in a dedicated right-hand panel, where input, output, and execution status are visible at a glance.
• Signed-in users opening the top page are no longer auto-redirected into the app until they click the login link, so the landing page and announcements stay browsable.
• The result panel and landing-page browsing path are separated so users can keep context while working.

Knowledge databases have been redesigned into an easy-to-read table grid with nine field types and inline schema editing from column headers. Files can now be previewed before upload.

• Redesigned knowledge databases into a table grid. Field types expanded to nine (adding long text, checkbox, single select, multiple select, URL, and email), and you can rename fields, change types, edit options, and delete columns directly from column headers.
• You can now preview a file's contents before confirming an upload. The file-selection display is also fully localized.
• Opening an empty folder now shows cards for creating a folder, a database, or a file, so even first-time users can start creating content without getting lost.

Added a Secret save request that lets AI safely guide you to the save screen instead of pasting sensitive values into chat. AI workers and the CLI can download knowledge files.

• Added a "Secret save request" so that instead of pasting sensitive values like API keys into chat, an AI agent safely guides you to the Secrets settings screen — opening a form in the right panel with the Secret name pre-filled.
• AI workers and the CLI can now download the contents of knowledge files. Deleting knowledge is now unified under full-edit permission; view/edit-share users cannot delete.
• The permission boundary was tightened so secret handling and knowledge-file access are safer to operate.

AI agents can now append or partially replace knowledge pages without loading them in full, browser-profile login sessions are auto-saved, and Slack multi-account routing no longer triggers duplicate runs.

• Added the `knowledge_append` and `knowledge_replace` lightweight knowledge editing tools so AI agents can append to or partially replace knowledge pages without loading them in full — making edits on large pages much faster.
• Browser-profile login sessions are now auto-saved without needing to press "Save and close". Sign in once before delegating to the AI and subsequent runs start from that logged-in state.
• When the same Slack workspace is linked to multiple Zemu accounts, a single Slack message no longer triggers multiple agents by mistake — Zemu now sends one disambiguation notice when several candidates match.

Scheduled triggers can now run as continuations of the same project thread, and you can create or update triggers right from chat with the new `/loop` command. Duplicate runs of the same trigger are skipped automatically.

• Scheduled triggers now support a "Use continuation project" switch. When enabled, each scheduled run continues as a follow-up to the same thread.
• You can now create or update scheduled triggers directly from chat by typing `/loop`. The AI confirms frequency, and when a specific time is given, it also detects the timezone.
• When a previous run of the same trigger is still in flight, the next run is now skipped automatically — duplicate concurrent executions no longer happen.

Knowledge access can now be granted by role, and AI agents can be embedded into external sites with a new public widget.

• Knowledge access can now be granted by role — view-only, append-only, or full edit — and public share links support issuance, revocation, and login-required settings.
• A new public embed widget lets you drop AI agents into your own site or service page. Configure allow-listed domains, the model used, and usage limits to publish safely.
• Editing Markdown tables inside knowledge has been refined — adding, removing, and reordering rows and columns is now much more intuitive.

AI workers can now keep browser login state via attached profiles, with new knowledge permission controls and the ability to hide past tasks.

• Attach a browser profile to an AI worker to persist login state and cookies during browser tool runs. A new agent tool also lets workers list, create, and assign profiles themselves.
• Knowledge permissions v2 adds shareable links, view-only access, and per-worker access grants so you can control exactly what each worker can read.
• Past tasks can now be hidden from the sidebar, with the hidden state synced per account across tabs.

Approve or reject AI worker actions directly from Slack, Discord, and LINE WORKS messages, without opening Zemu.

• Approve and reject buttons now appear in Slack, Discord, and LINE WORKS messages, so important approvals can be handled without leaving the chat.
• Error messages shown to users have been refined: internal technical details are hidden, and the wording is now easier to understand.
• Approval results now flow back from external chat, keeping team review workflows moving.

AI workers can now pause before selected tool executions so a person can approve or reject important actions.

• The new Permissions and Approvals settings let each worker control both allowed tools and tools that require pre-execution confirmation.
• When approval is required, the worker waits until a reviewer approves or rejects the request in the app, then resumes after approval.
• Worker runs are more resilient during backend updates, with up to 10 minutes to finish naturally instead of being interrupted immediately.

Native integrations are easier to recognize, and the AI thinking loader now has localized Japanese messages.

• Native integrations such as Slack, LINE WORKS, and Money Forward now show official service icons for easier recognition.
• The AI thinking loader now uses localized Japanese messages and clearer English copy while a response is being prepared.
• English loading messages were also refined so the response preparation state is easier to understand.

Slack Channel Route settings can now select private channels from the channel list.

• Slack Channel Route settings can now list private channels from the connected workspace.
• Private channels are labeled in the selector, so notification destinations can be configured without manually entering channel IDs.
• AI worker and trigger notifications are easier to route into internal private Slack channels.

A guided first-run tour now walks new users through the main product workflow.

• A seven-step tour now appears on the first dashboard visit, covering AI worker creation through the first chat.
• The tour explains external integrations, skills, triggers, and inbound channels while keeping users in the relevant product context.
• Users can choose a task-oriented chat starter at the end of the tour and continue directly into a new chat.

The Help Center is now organized by category, with additional UI improvements across navigation and worker setup.

• The Help Center was refreshed with Japanese and English guides for models, AI workers, skills, knowledge, triggers, integrations, billing, notifications, and member management.
• Settings now links directly to the Help Center, and the sidebar has a more compact, easier-to-scan layout.
• AI worker creation, worker icon editing, knowledge list display, and model selection guidance were refined.

New GPT models are available in the picker, making it easier to choose the right model for chats and AI workers.

• GPT-5.5 and GPT-5.4 are now available from the model selector.
• Native channel launches now default to GPT-5.4 for higher-quality responses from Slack and similar entry points.
• Model selection and default-model behavior are aligned across chats and external channels.

Manage plan selection, subscription changes, cancellations, and additional credit purchases from Billing settings.

• Monthly plans are shown in JPY before tax, with plan selection, changes, and cancellation managed from settings.
• Additional credits can be purchased for the amount you need. Purchased credits are valid for 120 days and are consumed oldest first.
• Plan lists were organized so current subscriptions and additional credit balances are easier to scan.

Connect Money Forward Cloud Accounting and Cloud Box separately, granting only the permissions needed for each workflow.

• Added Money Forward Cloud Box integration for evidence upload, listing, detail inspection, metadata updates, and download URL retrieval.
• Money Forward Accounting and Cloud Box are now separate connections, avoiding unnecessary permissions when only accounting access is needed.
• Money Forward profiles are identified by tenant name, and tenant metadata is retained after reauthorization.

Invoke AI workers from LINE WORKS and deliver trigger results to LINE WORKS channels.

• Connect via LINE WORKS Marketplace and invoke AI workers from group or 1:1 chats.
• Assign a LINE WORKS account to a worker and configure tools and a response policy (everyone / mention only / DM).
• Deliver schedule and event trigger results to LINE WORKS channels.

Filter skills by tag and author, and broadcast announcements to all users.

• Tag and author filters are now available on the skills list and in AI worker settings.
• Add, rename, or delete tags directly from the skill detail screen.
• New System Announcements feature lets administrators broadcast notices with Important and Light banner styles.

Adds an image generation and editing skill powered by OpenAI's latest GPT Image 2 model.

• New image generation and editing skill powered by OpenAI GPT Image 2.
• Use it for banners, presentations, OGP, article heroes, product visuals, and social posts via your AI workers.
• The skill can be assigned to workers from the built-in skills list.

Adds Moonshot AI's Kimi K2.6 to the model picker and refines skill insertion in the chat input.

• Moonshot AI's latest Kimi K2.6 model is available from the model selector.
• Selecting a skill from the chat input now preserves your in-progress text and inserts at the cursor.
• The update keeps model choice and prompt composition smoother in daily chat work.

Override notification destinations (Slack / Discord / email) on a per-trigger basis.

• Override notification destinations (Slack / Discord / email) on a per-trigger basis when creating or editing triggers.
• Triggers created from Slack or Discord auto-wire the source channel as their notification destination.
• Added a Slack thread lookup runtime tool and improved timezone inference when none is specified.

Reduces friction in initial AI worker setup and external tool connection.

• Set a worker name and visibility on the pre-creation screen and continue straight into settings.
• Trying to use an unactivated external tool now surfaces the required connection or setup, with smoother retry on connection errors.
• The setup flow now makes the next action clearer before the first worker run.

Rebuilds the spreadsheet editor and keeps workers running through transient external integration failures.

• Rebuilt spreadsheet editor for steadier CSV/XLSX editing, saving, and display.
• Workers continue past transient Slack or external integration failures and can surface follow-up guidance instead of stopping.
• The update improves day-to-day reliability for file work and connected workflows.

Discord server channels can now start AI workers, just like Slack.

• Start AI workers from Discord server channels through native channel settings.
• Connect with the official Zemu Discord app — no bot token entry required.
• Workers launched from Discord can use Discord-specific runtime tools for replies and file transfers.

Strengthens Slack-triggered workers around file handoff, progress replies, and thread binding.

• Workers can fetch Slack file attachments and send generated files back to Slack.
• The same Slack thread now holds separate Zemu projects per speaker, so user context stays isolated.
• Slack progress replies are paced by the AI, keeping status, completion, and failure messages concise.

Adds pricing, cases, release notes, and support pages, plus the Gemini image generation skill.

• Added marketing pages for pricing, cases, release notes, and support.
• Added a Gemini 3.1 Flash Image skill and an email reply agent starter.
• The public site now gives evaluators a clearer path to product information.

The zemu CLI now uploads local files directly to knowledge folders.

• zemu knowledge nodes upload registers local files into a knowledge folder.
• Script and automation pipelines can keep knowledge up to date without manual steps.
• CLI uploads appear alongside knowledge added from the web UI.

Creating a new organization now provisions the Zemu Customer Success worker immediately.

• Creating a new organization leaves the Zemu Customer Success worker ready to use.
• New teams can start from a ready-made customer success worker.
• The worker appears in the AI worker list immediately after organization setup.

The default worker "Zemu Customer Success" is built in, improving the initial experience.

• Zemu Customer Success is built in as the default system worker.
• A new chat with the default worker shows starter prompts for a quick first step.
• Opening a chat delivers a greeting from the AI worker, giving a clear cue to start.

A CLI-driven skill for knowledge DB operations and savable views for filters and sorts.

• Database Manager skill handles row add, update, delete, and view management for the knowledge DB.
• Save filter and sort conditions as views.
• Structured knowledge operations are easier to repeat and audit.

Pick a model per worker and use semantic search across knowledge.

• Select the model for each AI worker to match your workload.
• Knowledge search now supports semantic matching and is tolerant of phrasing differences.
• Teams can balance accuracy, cost, and retrieval quality per worker.

Save PDFs and images to knowledge, and switch between prompt, tools, and triggers tabs when configuring an AI worker.

• Store and manage PDFs, images, and other files as knowledge nodes.
• Worker settings are now tabbed, making it easy to switch between prompt, tools, and triggers.
• The settings tabs reduce friction when switching between worker configuration areas.

Added worker visibility controls and connection settings for combining workers.

• Choose each AI worker’s visibility: private, organization, or public.
• Configure which other workers an AI worker can call from the UI, making multi-agent workflows simpler to build.
• Teams can combine specialized workers while keeping usage scope controlled.

The agent builder UI is renewed, and skills can now be imported from GitHub.

• Redesigned the AI worker creation and configuration screens.
• Import a skill into Zemu by pointing at a GitHub repository URL.
• Existing workflow assets are easier to bring into Zemu.

Use-case-based AI worker templates and support for file attachments in chat.

• Stand up an AI worker quickly from use-case-based templates.
• Attach PDFs or images to a chat and let the AI worker use them as context.
• The update makes the first worker setup faster for common business tasks.

Export presentations, manage structured knowledge, and store API keys securely.

• Export AI-generated presentations to PPTX or PDF.
• Create database-shaped nodes in knowledge and manage structured data.
• Store secrets like API keys safely and reference them from skills.

Launch AI workers from external events and call workers from other workers.

• Automatically start AI workers from external events such as inbound email or Slack notifications.
• An AI worker can call another worker for multi-step workflows.
• Attach Composio connection profiles to agents for managed integrations.

Organize knowledge by folder, connect MCP servers, and run workers on a schedule.

• Organize knowledge in folders and create, edit, or delete pages directly from the UI.
• Connect MCP (Model Context Protocol) servers to AI workers.
• Automate recurring work with a schedule trigger that runs at specified times.

Fixed workflow skills, plan mode, shared-thread API, and self-prompt updates landed together.

• Define and run fixed processing flows as workflow skills.
• Invoke skills directly from plan mode or slash commands.
• Added a public API for shared threads and a mechanism for workers to update their own prompts.

Team account creation, thread pinning, Brave Search, and safer sandbox-to-external calls.

• Create team accounts directly from the sidebar's Organization menu.
• Pin important threads to the top of the sidebar so they are easy to find.
• Use Brave Search from the sandbox, and call external features safely from sandbox code.

Faster access to recent agents and conversation history from the sidebar.

• Jump straight back to a recently used agent from the chat sidebar.
• Conversation history navigation is organized for easier discovery.
• The sidebar changes make it easier to resume prior work.

Custom skills management UI was redesigned, and output guardrails were added.

• The skills management UI was redesigned, supporting uploads and GitHub imports.
• Added output guardrails that constrain what AI agents produce.
• The update improves both skill reuse and output control.

Initial availability for creating AI workers and requesting work from chat.

• Create, configure, and run AI workers
• Task execution starting from chat
• Organization-oriented management screens

Knowledge management for passing business context to AI workers.

• Register manuals and internal rules
• Reference knowledge during worker execution
• Manage content in a searchable structure

Paths for defining repeatable work and running it from schedules or events.

• Reuse fixed procedures as skills
• Automate from time-based or event-based triggers
• Keep semi-automated flows with approval