What’s new in Monad

We're making a focused change this week: Monad is removing its built-in task tracker. Tasks, task boards, #PREFIX-N task references, and personal task boards are going away.

What's changing

• Tasks and boards are gone. The board view of a canvas database stays — a kanban board view of your data is still one click away — but the standalone task tracker, task cards, and the per-user personal task board are removed.
• Old task references become plain text. A message that mentioned a task like #ENG-12 or linked a board keeps its wording, but it's now ordinary text rather than a clickable chip. Nothing breaks and nothing disappears from your history.
• Task links no longer open. Direct links to an individual task (the /task/… URLs) no longer resolve.

Where your work goes now

Monad is a communication, canvas, and agents workspace, and that's where we're putting our energy:

• Track work in a canvas database — add a status column, group by owner, and flip to a board view when you want a kanban layout.
• Keep plans and specs in canvas documents, right alongside the conversation.
• Hand recurring or multi-step work to an agent that lives in your workspace and acts on your behalf.

If your team relied on tasks, moving that work into a canvas database is the most direct replacement — and it comes with views, sharing, and agents your task boards never had.

A big one this week: a brand-new way to browse everything your team knows, plus channels that finally remember where you left off.

Meet Library

Library is a new sidebar surface, next to Search and Control Tower, that distills your workspace's conversations, files, and connected apps into living, grounded summaries — no more digging through old threads to reconstruct where a project or decision stands.

• Cards organize around the work that's actually happening in your workspace — projects, decisions, clients, candidates, matters, or whatever fits your business — and update automatically as new activity comes in.
• Every summary is grounded: open a card to see the exact messages, files, and connected-app content it's built from.
• Connect Gmail, Google Calendar, Google Drive, and X so Library can pull relevant context from those sources too, alongside your workspace conversations.
• Ask a question inside a cluster to get an answer sourced from that cluster's material.
• Follow a cluster to keep it on your radar, or pin sources you don't want to lose track of.
• Everything respects existing privacy boundaries — a private space, DM, or your personal Gmail never leaks into a workspace-wide summary.

Channels open where you left off

Clicking into a busy channel used to drop you at the very bottom, leaving you to scroll up and hunt for where you stopped reading. Now it opens centered on your first unread message instead, so you can pick up right where you left off. A channel you're already caught up on still opens instantly at the latest message.

Every channel also has a new ⋯ menu with Mark all as read and Mark as unread, so you can manage your read state deliberately whenever you want to.

Improvements

• Safer Gmail archiving. Archiving an email through an agent now removes it from your inbox without ever risking a move to Trash.
• Disconnecting a Google connector now revokes access at Google, too — not just inside Monad, so a disconnected app no longer lingers in your Google account's list of connected apps.
• My Connections is easier to manage. Connections that need reconnecting can now be removed directly, without reconnecting first, and each connection's tool list groups read-only tools ahead of the ones that can make changes.
• Smoother channel switching. Jumping between channels no longer briefly flashes plain, unstyled text before the full message view loads.
• Clearer sync recovery messaging. When sync pauses, the banner now reads "Sync is paused — restart to reconnect" with a consistent Restart sync action.
• Uploads are now scanned for malware before they're stored, so a malicious file can't be shared through your workspace.
• New account passwords must now be at least 12 characters.

A quieter week on the surface — under the hood the team is heads-down on bigger features landing soon. The one visible change: connecting Gmail just got simpler.

A cleaner consent screen for Gmail

When you connect your Gmail account, the OAuth consent screen now asks for a single permission instead of four. Previously Monad requested separate permissions for reading, composing, labeling, and modifying mail — but one of those already covers everything the other three do. The result is a shorter, easier-to-understand grant that does exactly the same thing.

If you already connected Gmail, nothing changes — your existing connection keeps working as before. The simpler screen only appears for new connections.

Quiet a noisy space without suppressing source facts or agent work. Control Tower cards now tell you what an agent actually did. And share or archive any canvas without hunting for the tab right-click menu.

Space attention and Herald

Each channel now has one attention setting: All new posts, Just mentions, or Mute. It controls human awareness and outward delivery without deleting the underlying Loom fact or suppressing authorized agent activation. An independent hard path such as a direct mention can still win.

Under Settings → Notifications, Herald preferences control outward delivery, sound, digests, quiet hours, caps, and delivery-class overrides.

Agent oversight cards

When an agent acts on your behalf — sending a message, updating a task, joining a space — you'll now see a distinct oversight card in Control Tower with a concrete description of what happened ("Sent a message to #design" instead of a generic tool name). Actions taken through connected apps appear there too, so your oversight covers everything the agent does.

When an agent needs your input to continue, the card gets a clear visual accent and an Answer button so you know exactly where your response is needed.

Share and archive from canvas menus

Share to space and archive actions are now available directly from the triple-dot menu inside any canvas database or canvas document. Previously these were only reachable by right-clicking the canvas tab in the sidebar, which is easy to miss while you're focused on your work.

The menu adapts based on context: if a canvas lives in only one space, you'll see Archive; if it's shared across multiple spaces, you'll see Remove from space instead — so you can never accidentally archive a canvas that's still being used elsewhere.

Improvements

• Agents now respond to workspace facts across all supported event types. Database edits, task assignments, and comment resolutions wake subscribed agents through Loom, independently of Herald.

Fixes

• Direct mentions remain independently important. Muting ambient channel traffic cannot erase the direct-targeting evidence path.
• Canvas page menu renders correctly for hosts with partial callbacks. The archive/remove separator no longer appears when the host only provides one lifecycle callback.

A big week: agent billing moves to a new model, Control Tower separates obligations from awareness, and agents proactively act on attachments you drop in a space.

Agent billing by the hour

Agent runtime now bills by the hour, not by the token. Three tiers, three rates:

• Frontier — $10/hr. The highest-capability models for complex reasoning, writing, and research.
• Balanced — $4/hr. Strong all-around models for day-to-day tasks.
• Cost-Effective — $2/hr. Fast, lightweight models for high-volume, lower-complexity work.

Do and Know in Control Tower

Control Tower now makes the product distinction explicit. Do holds work that needs an outcome and is not cleared by reading; Know assembles grounded awareness into an edition you acknowledge as a whole.

Three details make the model reliable:

• Recipient-relative ranking. The same message can matter directly to you and have different relevance through several Library topics without those assessments overwriting each other.
• Generation-scoped obligations. A retraction closes only the current ask; a genuinely new ask on the same topic can appear again.
• Exact unread discharge. Know clears from positioned surface progress, not timestamps or lifetime group totals.

Agents that respond to file drops

Drop a PDF, Word doc, spreadsheet, or text file into any space and the right agent activates. Monad now reads the file name, type, and a preview of its contents (this was fun to build!) before deciding which agent(s) should respond — so it knows what you dropped without you having to explain it.

Improvements

• Delete canvas database views. Right-click any view tab inside a canvas database to delete it. A confirmation dialog shows the view name before anything is removed. The last remaining view on a database can't be deleted — at least one always stays.
• Onboarding confirms what Monad found about your company first. Before the setup wizard begins, you see a read-only summary of what Monad researched and can confirm it looks right — so your workspace is set up around the right company from the start.
• Plan editor requires confirmation before deleting. Remove buttons in the workspace plan editor are always visible (no longer hover-only), and all removals — plan rows and pending invites — need confirmation before they take effect.
• Company research no longer delays attention delivery. Long-running research tasks now run separately, so Control Tower, Herald, and search stay fast even while Monad is looking something up.

Fixes

• Agents now retry when a response is cut off mid-stream. Truncated AI responses previously failed silently. Monad detects the truncation and retries automatically.
• Space names are cleaned up automatically. If Monad proposes a space called #general, it's saved as general — the leading # is stripped.
• Onboarding won't proceed with an incomplete plan. A built-in check requires the proposed workspace plan to be confirmed complete before setup can continue.

Five things shipped this week that each change how you use Monad every day — from the first screen you see when you open the app to how your agents reach your inbox on a schedule.

Control Tower

Your workspace home is now Control Tower, split around two different jobs: Do keeps obligations open until their outcome resolves, while Know is an immutable, grounded edition of what is worth catching up on.

• Live obligations: decisions, approvals, reviews, replies, agent input, and assigned tasks stay visible even after you open their source.
• Grounded awareness: Know freezes exact source revisions and citations, with deterministic fallback if richer compilation is unavailable.
• Current links: every card resolves its current authorized source rather than storing a stale URL.
• Offline-safe actions: completion and dismissal use generation checks, so a stale tab cannot complete a replacement obligation.

Connections — your tools, inside Monad

The workspace has a new Connections gallery (formerly "MCP servers") where you link external services to your agents in one click. Connect once and every agent you've given access can act through that tool on your behalf.

This week's additions:

• Gmail, Google Calendar, Google Drive: click Connect, authorize with your Google account, and you're done — no app to configure, no credentials to find. Agents can read and send email, manage calendar events, and read Drive files.
• Notion and Linear: same one-click experience. Agents can browse Notion pages and databases, and read and create Linear issues.
• X: search, post, and monitor mentions. X account connection was broken last week — that's fixed, and the connector now gains Lists support and enriched tweet data. Reconnect once from Settings → Connections → X to pick up Lists permissions.
• Personal connections: any member can install a connector privately without waiting for an admin.

Monad is everywhere you are

The Monad agent is now omnipresent — pinned at the top of your DMs, reachable with @Monad from any space, and aware of who you are wherever it shows up. It carries per-user memory across conversations and shows only you what you share with it.

Start a thought in a channel, continue it in DMs. Your coworkers see the trigger and Monad's reply, but not the private context between you. Every interaction is anchored to you as the caller.

Recurring schedules

You can now give Monad a standing instruction and let it run on a cadence. Define what you want — a daily inbox triage, a weekly project summary, a recurring competitor check — and Monad wakes on that schedule, runs as you (with your connections and permissions), and DMs you the result.

Each schedule requires a one-time consent grant you can revoke anytime.

Thread replies that reach the channel

When you reply to a thread, you can now also surface that reply in the parent channel. Check the "Also send to #channel" box under the reply composer and your message appears in the channel feed as well as in the thread — one message, visible in both places, counting toward unread and triggering notifications.

Improvements

• Settings → Connections: the "MCP servers" page is now called "Connections," with a gallery view and a dedicated detail page for each service. Old bookmarks redirect automatically.
• Real brand logos: connectors show their actual icons instead of a generic first-letter placeholder.
• @mention warning that sticks: when you @mention someone not in the channel, a persistent note (visible only to you) appears with a one-click "Add to channel" button. It survives page reloads, unlike the old banner.
• Personal email onboarding: if you sign up with a Gmail or Outlook address, you'll see an optional Work Website field. Monad uses it to personalize your setup experience.
• New Monad logo mark: updated brand identity (rounded-square cluster) is now live across the app icon, browser tab, share previews, and the desktop app.
• Personal MCP servers: members can add remote-MCP connectors as private personal servers without needing admin action.
• Desktop app v0.1.0: after updating to a new code-signing certificate, the desktop app is back on a stable release track.

Fixes

• Notion and Linear connections failed on setup: every connection attempt immediately showed an error. Fixed.
• X account connection always failed: connecting your X account failed every time. Fixed.
• Gmail returned a permission error on every call: Gmail requests were rejected for all accounts. Fixed — Gmail now works reliably with any connected Google account.
• X search returned errors on every query: search queries were silently failing. Fixed.
• Notion tools failed on the default agent: Notion actions failed with confusing errors for some agent configurations. Fixed — compatibility improvements now apply across all connectors.
• Agent runs crashed instead of recovering: when a request was rejected, the entire run ended. Fixed — runs now retry automatically instead of stopping.
• Desktop app crashed on first launch after updating to a new code-signing identity. Fixed.

A week packed with new ways to bring people in, tune what you hear, and connect your tools, all without leaving Monad.

Guest invite links

You can now create a reusable guest invite link with no individual invites to send. Share the link in an email, a doc, or a client-facing chat and guests land directly in the channels you've scoped for them.

• Multi-use and shareable: one link works for everyone you send it to; revoke it anytime from Settings → Guest Links.
• Channel-scoped: guests only see the spaces you include on the link, keeping the rest of your workspace private.
• Pending approval: admins get a notification for each new claim and can approve or reject before the guest is fully admitted.

Per-space notification preferences

Every space now has its own notification level, accessible from the bell icon in the space header. Dial down a high-volume channel to mentions-only without touching your global settings.

• Three levels: All new posts, Just mentions, or Mute.
• Lives in the header: click the bell next to the space name to change it instantly.
• Synced across devices: your preference follows you on desktop, mobile, and the web app.

Reaction notifications

When someone adds an emoji reaction to one of your messages, you now get a notification, so you actually know the message landed.

MCP tools connect automatically

Connecting an MCP server used to mean copying a client ID, pasting a secret, and navigating an OAuth flow by hand. Now Monad discovers the right settings automatically via OAuth Dynamic Client Registration. Paste the server URL and click Connect, and Monad handles the rest.

• Zero copy-paste: no client ID or secret to manage.
• Works with any standards-compliant MCP server: if the server supports dynamic registration, Monad auto-configures it.

Prepaid credits

You can now load your workspace with a credit balance and let it draw down as you use AI features. Set up auto-recharge to top up automatically when your balance drops below a threshold, or manage it manually from Settings → Credits.

• Monthly grants: plans include a monthly credit allocation at no extra charge.
• Auto-recharge: set a low-balance threshold and a top-up amount so you never hit zero mid-workflow.
• Transparent usage: see a live balance and a per-day breakdown right in your settings.

Improvements

• Archived channels: spaces you've archived disappear from the sidebar and channel browser. A banner on the archived channel lets anyone with access reopen it.
• Space header face-pile: see who's in a space at a glance; the member count and faces now appear in the header alongside the notification bell.
• Onboarding wizard: responses stream in as the agent types, the wizard now suggests context-aware reply chips, and you can schedule recurring agent check-ins right from the setup chat.
• Collapsed join events: back-to-back join messages in a channel are now grouped into a single compact line so the conversation stays readable.
• iOS home screen: the installed app caches key resources more aggressively, making cold launches faster on iPhone and iPad.
• Workspace deletion: owners can now permanently delete a workspace from Settings → Workspace. Deletion is durable and fully tears down all associated data.
• Add members modal: adding people to a space now opens a focused search dialog instead of navigating away.
• Agents load skills more eagerly: when it's ambiguous whether a skill is relevant, agents now lean toward loading it so they stay helpful across more conversations.
• Sharper agent memory: the nightly memory-tidying pass now uses a more capable model to curate what agents remember, improving consistency across long conversations. Every agent now dreams for each person it works with, and a new dashboard lets you review, approve, or roll back any memory changes before they take effect.

Fixes

• @-mention popup: the suggestion menu now anchors to your cursor position instead of snapping to the editor corner.
• Canvas documents: documents that were blank on first load now render instantly using a locally-cached snapshot.
• Mention chip avatars: avatars and emoji inside collapsed mention chips now display correctly in all rich-text surfaces.
• Message composer: fixed several glitches when splitting lines, creating code blocks from selected text, and applying formatting shortcuts.
• Admin session navigation: clicking a row in the agent sessions list now opens the correct session detail.

This week we’re giving your digital coworkers better memory, the ability to learn new procedures on the fly, and a way to improve their own performance based on how you interact with them.

Agent skills

You can now attach reusable modules of knowledge and procedures to any agent. Instead of writing long, repetitive instructions for every agent, you can define skills once and let your agents load them when they need them.

• Procedural playbooks — Define "how we file a bug" or "our brand voice" as a skill shared across your team.
• On-demand activation — Agents only load the full text of a skill when it becomes relevant, keeping their conversations focused and efficient.
• Workspace authored — Create your own skills in Settings → Skills to capture your team's unique domain knowledge.

Self-improving agents

Your agents now learn from your feedback. When you give a thumbs-up or thumbs-down to an agent's response, Monad analyzes the gap between the actual response and what you expected.

• Instruction refinement — The system periodically optimizes an agent's internal instructions to improve its accuracy.
• Safety first — Every proposed improvement is measured against a history of your past feedback to ensure there are no regressions before it goes live.
• Transparency — You can review, approve, or revert any auto-generated instruction changes in the Agent Optimization dashboard.

Agents now dream

To keep your long-term conversations coherent, agents now perform a nightly memory consolidation pass. We call this "dreaming."

• Coherent history — Agents deduplicate, merge, and prune their stored memories overnight, resolving contradictions and tidying up stale facts.
• Contextual relevance — By tidying their own notes while you sleep, agents stay focused on what matters most to your team during the day.

Virtual model tiers

We’ve introduced three virtual model tiers—Frontier, Balanced, and Cost-effective—to give you more control over the balance between intelligence and cost.

• Frontier — The most capable models for complex reasoning and creative tasks.
• Balanced — A blend of speed and intelligence for your everyday workflows.
• Cost-effective — Fast, lightweight models for high-volume tasks like basic summarization.

Improvements

• Admin identity — Team admins can now access a new platform portal with support for federated identity (OIDC).
• Onboarding polish — A smoother setup experience with better message tracking and clearer navigation.
• Legal suite — Completely updated Privacy, Terms, and Data Processing policies grounded in our commitment to transparency.
• Refined feedback — Better UI for rating agent responses and seeing the impact on their instructions.

Fixes

• Browser reliability — Fixed an authentication loop that affected some Safari users.
• Integration reliability — Resolved an issue where some third-party tools would fail to load on high-performance models.
• UI Polish — Fixed sidebar flickering and tooltip overlapping issues during heavy message activity.

A packed end-of-week push: fix your messages after you send them, take Monad with you on iPhone, and find anything across your entire workspace.

Edit messages and comments

Sent something with a typo — or just changed your mind? You can now edit your messages for up to 24 hours after posting. The edit opens right in place, pre-filled with what you wrote.

• Change the text, swap out attachments, update mentions — the full composer, in the message.
• Press Esc or click Cancel to back out without saving.
• An "Edited" label appears once you save, visible to everyone live — no refresh needed.
• Editing works the same way on canvas document comments and task comments: open the menu, choose Edit, and save.
• If you add a new mention in an edit, those people get notified — but people already mentioned aren't pinged again.

Monad is on iPhone and iPad

Download Monad from the App Store and you get the full web experience wrapped in a native shell — push notifications, deep links from emails and messages, and an offline screen when you lose signal.

Search everything

Search now reaches across every canvas database, board, and task in your workspace — not just messages and documents.

• Find a row in a database, a card on a board, or a task by name, description, or any text it contains.
• Type # in the composer to pull up board suggestions for the current space, prefetched so they appear immediately.

Canvas database headers

Every canvas database now has an editable header: set a name, pick an icon, and write a short description. Changes stream live to everyone viewing the database — no refresh required.

MCP tool control

If your workspace connects to external MCP servers, you can now disable individual tools without disconnecting the whole server. Toggle a tool off in MCP server settings and it disappears from every agent that uses that server — instantly, mid-run included.

macOS desktop: push notifications

The macOS desktop app now supports push notifications via Apple Push Notifications. Turn them on from notification settings, and your unread badge updates in the background even when the window is closed.

Improvements

• Canvas icon and cover changes stream live to everyone viewing the canvas — the tab bar and page header stay in sync automatically.
• The agent session viewer now collapses older tool-call rounds so you can focus on what's happening now, and tool-call steps in the current round are collapsed by default.
• Agents can seed initial content into canvas documents when they create them.
• The Agents section in the sidebar is now collapsed by default to give your channels and DMs more room.
• Thread reply notifications now deep-link straight to the reply in context.
• Workspace logos appear in the desktop sign-in picker.

Fixes

• Fixed dead-clicks in message lists when rows were still loading.
• Fixed a double border that appeared on canvas spreadsheet cells while editing.
• Group DM avatars now render a gradient fallback and legible initials.
• Monad no longer auto-opens the thread panel when an agent or user triggers a thread in a DM conversation.

A big week: a whole new way to react, exact conversation read state, more capable agents, and Monad on your home screen.

Custom emoji

Make the workspace yours. Upload your own emoji and use them anywhere you'd use a regular one — as reactions and right inside your messages.

• Type : to open a picker with 1,400+ built-in emoji plus everything your workspace has added, and search by name.
• Give an emoji more than one name with aliases.
• Drop in a jumbo emoji when one just isn't enough.
• Admins decide whether everyone can add emoji or just admins.

Exact read progress in threads

Threads and channel feeds now track the exact committed occurrence you have seen, independent of timestamps. A reply that lands while you are opening a thread stays new, and a delayed message cannot disappear behind an older clock.

• Thread roots and replies have stable positions, so multi-device acknowledgements only move forward.
• Joining a channel starts at the committed head; old history does not suddenly become unread.
• Crossposting an older message creates a new destination occurrence, so it can correctly be new there.
• Push settings and test delivery remain under notification preferences.

Rate your agents

Every agent reply now has a thumbs up and thumbs down. Add a quick reason when something misses, and your feedback helps that agent get better over time.

Agents are also more proactive: they read along in the channels they belong to and step into relevant conversations on their own, instead of waiting to be asked every time.

Install Monad as an app

Monad is now installable as an app on your desktop and phone. Add it to your home screen or dock, and your unread count shows up as a badge on the icon — even when the window is closed.

Canvas databases, leveled up

Working with structured data got a lot smoother:

• Edit columns inline and pick from select options without leaving the cell.
• Drag columns to reorder them, and open a dedicated panel to edit a column's type and settings.
• Export any database to CSV.
• Scroll through large tables with smooth, automatic pagination.

Connect Monad over MCP

Monad now exposes a curated set of tools over the Model Context Protocol. Point an MCP-compatible client at your workspace, authorize it, and read and act on your workspace from outside Monad.

Improvements

• Paste a Monad link into the composer and it turns into a mention chip automatically.
• Mention people and agents inside threads, and in canvas documents.
• Discover and join spaces from a new space browser.
• Set your workspace name and logo from the new workspace settings page.
• Pre-select spaces and groups when you invite new teammates.
• Recognized sign-in providers now show their branded avatars.
• Messages show where they were sent from — web, mobile, the API, or an agent.
• GitHub-style task lists and tables now render in messages and documents.
• Realtime surfaces refresh the moment you reconnect after going offline.

Fixes

• Removing the last reaction on a message now updates live for everyone.
• Drag-and-drop overlays no longer get stuck on screen after a cancelled drag.
• Fixed horizontal overflow in thread panels on mobile.
• A round of canvas table layout and editing fixes.

This week is about polish where you spend the most time — writing documents and moving work across your boards.

A more polished document editor

Writing in Monad feels noticeably smoother:

• Inline autocomplete now understands more context and lets you cycle through suggestions before you accept one.
• Paste mixed content — multiple block types and nested lists at once — and it lands intact instead of getting flattened.
• List spacing is cleaner, with the cramped gaps between list items gone.
• Comments stay in sync as you edit, resolved threads no longer leave stray highlights behind, and you're warned when an @mention in a comment can't be delivered.
• Mention chips and document titles stay consistent across every session and device.

Task boards that do more

Boards picked up the things you'd actually use day to day:

• Comment on a task right from a cleaner task detail view.
• Drag a card to position it exactly where you want within a column.
• Filter board columns by a time window to focus on what's recent.
• Reorder the tabs across the top of a space.
• Manage sub-tasks from a tidy dropdown on the parent card.

Mention boards anywhere

The # menu now includes boards, so you can link to a board the same way you link to a channel or document.

A smarter model picker

Choosing which model an agent uses is faster and clearer. Models are grouped by the lab that makes them and ranked by a refreshed intelligence-and-value score, trimmed to the strongest options so you're not scrolling through a long list. You can also explicitly choose no model where it makes sense.

Improvements

• Stacked banners now size themselves correctly instead of overlapping.
• Group context carries through as you navigate between spaces.

Fixes

• Fixed author coloring being inherited incorrectly while editing together.
• Fixed list structure corruption detection and repair.
• Fixed a brief disconnect that could happen while the editor was starting up.
• Streamlined agent replies by removing the typing indicator.