Release notes

Cloud PDFs open reliably

PDFs uploaded to a cloud workspace now open consistently in both the desktop app and the web app. Documents with mixed page sizes and multilingual fonts also render with more accurate layout.

Workspace switching feels instant

Consilience securely remembers the cloud workspaces for the signed-in account, so the switcher can show them as soon as it opens. The list refreshes in the background, while folder expansion reuses the current file tree to avoid repeated waits.

A lighter path to the first screen

The workspace shell now appears first, while the editor, settings, ingest, connections, and feeds load only when needed. This reduces the initial core code by about 10% and improves first-screen responsiveness across desktop and web.

One Consilience, wherever your work lives

Cloud workspaces now open directly inside the complete Consilience desktop experience. Switch between local and cloud workspaces in the same window, with the same sidebar, tabs, editor, search, graph, and agent tools. Authentication still uses your default browser when it is actually needed, then returns you to the app.

Files that behave like local files

Cloud files now support the everyday workflows you expect from a local workspace: drag images and documents into notes, drop files or nested folders into the file tree, rename, copy, move, duplicate, restore from Trash, and download a portable workspace copy. PDF viewing, images, and design assets also stay inside the familiar app flow.

More dependable hosted work

Search, knowledge graphs, conversations, research, and workspace intelligence now share a more consistent hosted experience. This release also improves name-conflict handling and keeps unavailable editing controls out of view for read-only members.

Cloud workspaces open on the desktop

Opening a cloud workspace from the desktop app used to give you an empty window. It now opens the workspace, on both macOS and Windows.

That window behaves like a window, too. It closes when you ask it to, and the menu items and keyboard shortcuts you use inside it act on the window you are actually looking at instead of quietly acting on another one. Separately, a window that was still reading your notes could refuse to close at all, with no way out short of quitting; it now closes when you tell it to, and says so if it cannot.

The hosted app caught up with the desktop

You can now bring sources into a hosted workspace by pasting a link or pasting text, without switching machines. Chat there finally receives what you gave it: an attached image, the mode you picked, and the note you have open all reach the assistant now, so answers stop being confidently uninformed. The deeper run modes work as well, not just the quick ones.

More of the workspace is real rather than reported as empty. Backlinks show the notes that actually link to a note instead of a flat zero, cloud workspaces can be renamed and deleted from the workspace menu, and the topic map is visible in the browser, with the owner able to rebuild it. File rows also wear their proper icons.

The browser opens much sooner

The hosted app starts noticeably faster. It now paints the workspace while it finishes signing you in rather than after, downloads only the language you are actually reading, and no longer loads the editor or the map before you have opened either one.

Your chats stay with their workspace

On Windows, a workspace's chats could vanish after an update. They were never deleted; the app had started looking for them under a different name for the same folder. This release finds them again on first launch, and makes the two spellings impossible to tell apart from now on.

Closing a workspace also used to delete every chat in it, without warning and one click away in the sidebar. It no longer does. Closing is now only closing, and reopening the folder brings the conversations back with it.

Starting a workspace, and getting documents in

You can create a new local workspace from the workspace menu and from the first-run screen. Until now every way in led to a folder that had to already exist, which made a new workspace something you had to make somewhere else first.

Importing documents is steadier and cheaper. A page that comes back blank is now retried instead of failing the whole source, and reading scanned pages costs meaningfully fewer credits than it did.

Your whole workspace now opens in the browser

A workspace published from the desktop used to arrive in the browser with only its notes. PDFs, images, spreadsheets, source files and design canvases were simply absent, with no row and no message to say the view was partial, which reads less like "this is a partial view" and more like "my files are gone."

The browser now shows every file in the workspace, and the text ones open and save with the same size limits and the same safeguards the desktop editor uses. Opening a source file also used to replace the entire app with an error screen; that is fixed, and the panel behind it no longer appears where it could never have had anything to show.

Workspaces get a real name, in any language

Naming a workspace in Korean, Japanese or Chinese used to leave you with exactly one workspace. The first one worked and every one after it failed, whatever you called it. A name with a space in it failed outright, and all you were told was that something had returned an error.

A workspace now has a name you choose, kept separate from the address that points at it. Any language works, two workspaces may share a name the way two folders can, and the name you typed is the name you see everywhere you manage workspaces. When a workspace cannot be created, you are told the actual reason.

Every note has an address you can share

The hosted app lived at one address no matter what you were reading. You could not send a colleague a link to a note, pasting the address into a second tab gave you an empty workspace, and the browser's Back button left the app entirely.

The note you are reading now names itself in the address bar. A link can be shared, a reload returns you to what you were reading, and Back walks back through the notes you opened instead of closing the app.

Comfortable on a narrow screen

At phone width the editor was squeezed down to almost nothing, with room for the line numbers and little else. Below a certain width the side panel now steps out of the layout and opens over it as a drawer, so the note keeps the full width of the window.

The drawer also stays out of the way when you arrive straight at a document, from a shared link or a restored tab, instead of opening on top of the thing you came to read.

The web app stops offering what it cannot do

The largest category of fixes in this release is not missing features but convincing ones. Controls the browser could not honour still rendered, and panels that could not read your workspace drew their empty state instead of an error, so "this workspace has nothing" and "this was never built here" looked identical.

The workspace status panel was the worst of them: it reported that everything had been processed, about a workspace it had not managed to read at all. The import wizard ran all five of its stages before failing. Commands that could only ever fail were still offered, and turned you down only after you had written the whole request. Those controls are now either gone or honest, and when something is unavailable it says so in your own language rather than in English.

A faster first visit, and three things that quietly did not work

The first visit of the day to the hosted app used to sit for a long, unexplained pause before anything appeared. That pause is gone, and a first visit now opens at the speed of a return visit.

Three more things had only ever worked from the desktop. A purchase started in the browser now actually completes instead of ending nowhere. The map renders. And the people list on a shared workspace loads, so invitations, role changes and removals can be handled from the browser as they are on the desktop. The assistant also stops offering tools a hosted session does not have, which is what made it occasionally refuse a request it should have been able to handle.

Hosted workspaces now open where you expect

The hosted app now takes you straight into your workspace after sign-in, with the right product branding and app icon throughout. Switching workspaces from the titlebar now actually changes workspaces, duplicate entries have been removed, and old or unavailable workspaces are no longer shown as if they can be opened.

We also cleaned up hosted-only rough edges: unavailable desktop features are hidden or clearly marked instead of appearing as buttons that do nothing, and several screens that could get stuck loading or crash now open cleanly.

More reliable editing and saving in the browser

Saving notes in the hosted app is more dependable. Large notes no longer report a failed save after the content has already been written, manual saves no longer create conflicts with your own tab, and two tabs creating notes at the same time will no longer overwrite each other.

Typing while a note is still aligning with the server is also safer now; text entered during that moment is preserved instead of being replaced.

Images in hosted notes display correctly

Images attached to a note now load properly in the browser, including images referenced from within the note itself. Remote images in notes are also handled more reliably, so published and hosted workspaces look closer to what you see while editing.

Clearer language across the app

Several interface strings that were falling back to English now appear in the selected language, including parts of the workspace menu and sign-in flow. The sign-in screen also now identifies its language correctly for browsers and assistive technology.

Assistant-facing cards and prompts now use the right language more consistently, and a few labels and settings descriptions were tightened so they better match what the app actually shows.

Cloud workspaces in the browser

Consilience now supports hosted cloud workspaces alongside local workspaces. You can sign in, create or open a workspace from the web, and use the core app experience in a browser: chat, notes, file operations, graph views, entity pages, review flows, and workspace search are now available without opening the desktop app.

New accounts start with a ready-to-explore workspace instead of an empty picker, and the account console now puts your workspaces first. You can open workspaces directly, see recent activity, manage access, and download or restore workspace content more easily.

Shared workspaces with roles, invites, and live editing

Hosted workspaces can now be shared with other people. Owners can invite collaborators, assign roles, and send invitations even before someone has created an account. Sign-in now preserves invite links correctly, so invited members land in the workspace they were trying to join.

The web editor now supports live collaborative editing with visible presence, safer conflict handling, and clearer protection against accidental overwrites. Viewers remain read-only, editors can contribute, and activity history gives teams a readable record of what changed.

A simpler desktop shell

The desktop app has a redesigned workspace shell that keeps navigation closer to your work. The titlebar now shows the current workspace and lets you switch between recent, local, cloud, and shared workspaces; the sidebar now carries search, graph, activity, and workspace navigation in one consistent place.

Consilience now reopens your most recent workspace on launch, retires the old welcome screen, and gives empty workspaces clearer next steps. Sharing has also been consolidated into one workspace menu, so exporting, publishing, and inviting no longer feel like separate paths.

Clearer control over extraction, cost, and privacy

Extraction is now more explicit. Saving a note no longer has to mean extracting from it, and actions that may create usage now ask first and show an estimated cost before they run. Re-extraction requests, hosted saves, and account limits now present clearer messages instead of surprising background work.

This release also tightens several privacy and reliability edges around hosted workspaces, publishing, imports, PDF handling, backups, deletion, and account metering. In practice, that means fewer silent failures, better recovery from interruptions, more accurate usage records, and stronger protection for private note content.

A smoother graph review queue

Graph review is now more predictable and easier to work through. Suggested links and possible mistaken merges keep stable names and ordering while you review them, counters update correctly, and graph tuning panels refresh when you return to them.

You can also review combined identities from the existing Synonyms area, see the items involved, and separate them when needed without jumping between duplicate review surfaces.

More dependable imports and captures

Running multiple imports at once no longer mixes progress or results between batches. Web captures now handle inaccessible pages more safely instead of saving error pages as notes, and certain unusual link text no longer interrupts extraction.

Feed removal is also more reliable, including cases where previously saved items overlap with items from other feeds.

Search results that better match intent

Note search has been adjusted to produce more useful ordering across mixed queries. Results should feel less dominated by narrow matches and more likely to surface notes that match what you were actually looking for.

Settings, notifications, and editing fixes

The “Reopen AI sidebar on launch” setting now takes effect, and your default AI model choice is remembered after restarting Consilience. The notifications menu opens much faster, and clearing notifications now clears the badge as expected.

Consilience also now preserves your answer when a file changes on disk, improves localization coverage for work-loss dialogs, and displays Korean text correctly in background task output.

More ways to view your knowledge graph

Consilience now gives you several new ways to explore the shape of your workspace, including galaxy, shells, helix, and constellation-style views alongside the classic globe. Switching views feels smoother, with the camera reframing first so each new layout lands cleanly in view.

You can also switch graph views from the keyboard with a new customizable shortcut, making it easier to move between perspectives while you work.

A more dependable AI assistant

The AI assistant is more resilient during long or complex turns. Temporary connection issues are less likely to end a response, stalled requests are handled more gracefully, and failures are surfaced more clearly instead of leaving you waiting without feedback.

Stopping an AI action is also more reliable, including when the assistant is in the middle of a running task. Behind the scenes, the default premium AI experience has been refreshed so new conversations use the latest available option automatically.

Richer feeds and price history

Feed items now show clearer source identity with per-item publisher icons, making it easier to scan where stories are coming from at a glance. Price cards also include deeper history and a larger chart view for better context.

Longer chart ranges such as 3M, 1Y, and MAX now load the appropriate historical depth instead of showing the same window as shorter ranges, and range labels better follow your locale.

A clearer starting point

The Welcome screen has been redesigned with faster access to open, clone, and import actions, a cleaner first-run experience, and a new Academy entry point in place of the previous tour. You can also jump into key areas like feeds, cases, and research from the main workspace, while source control now lives in the workspace menu where it’s easier to find when needed.

Globe view is now the default graph experience

The knowledge graph now opens in the globe view by default, giving themes a clearer sense of place and making dense vaults easier to scan at a glance. You can still switch back to the classic layout from the view settings.

Inspecting a node is more consistent across graph modes: selecting an item brings its surrounding connections into focus, and entity pages now show a larger, resizable neighborhood view with more of the relevant context. We also fixed cases where saved graph layouts could reopen in an overly rigid or distorted arrangement.

Live feeds are easier to understand and control

Live feeds now make it clearer when something is waiting for your attention. The feed indicator shows pending activity, and the first time you enable live updates, Consilience asks how often you want it to check instead of silently choosing a schedule.

Feed runs are also better at respecting your vault. Consilience is more careful about avoiding duplicate saved stories and gives you a more predictable way to keep sources current on your own cadence.

Knowledge tuning gives clearer guidance

The knowledge tuning experience is more unified across the app. Status messages, recommendations, and one-click actions now point to the same next step, so it’s easier to understand what your vault needs and what will happen when you act.

We also improved error handling and recovery. Failed loads, stalled runs, and blocked actions now show as real issues instead of looking like empty or finished states. Theme naming is more likely to match the language of your notes, and problematic naming batches can be corrected more easily.

More reliable navigation and cleaner Windows polish

On Windows, Consilience now handles workspace paths more consistently, preventing cases where the same file could appear to be two different items. Report links from chat now open the intended file instead of occasionally taking you to an unrelated entity page.

The Windows interface also feels cleaner, with slimmer auto-hiding scrollbars and fewer nested scrolling areas in entity signal lists. Waiting and empty states now use the app’s new animated orb consistently, making background activity easier to recognize without adding clutter.

Vault-specific vocabulary and facts

Consilience can now adapt the knowledge graph to the way each vault thinks. You can define your own entity kinds and structured facts, so a workspace can track things like dates, prices, headcounts, or domain-specific labels instead of forcing everything into broad built-in categories.

The vocabulary experience has also been consolidated into one clearer editing flow, with suggestions based on what your vault already contains. Terms are kept in the language of your notes, so multilingual workspaces feel more natural and consistent.

Photos that connect to entities

Photos are now first-class context in the knowledge graph. Entity pages can show the images connected to that entity, and you can confirm which photos are actually of it. This also fixes several cases where imported screenshots or images with spaces in their filenames were present in the vault but not appearing where expected.

You can also find photos by describing what they show, even if you do not remember the file name. Opening photos from the graph now works reliably, and new vaults can attach their first image to an entity without needing an existing suggestion first.

Steadier entity pages and better sources

Entity pages now stay visually stable while live updates arrive. Sections should no longer flicker, jump, or overlap as new notes, feeds, and graph updates come in.

Feed handling is also more precise. Turning off News no longer removes curated sources you chose yourself, and the source curator is better at avoiding broad, noisy publication feeds that can fill an entity page with unrelated articles.

More reliable graph automation

Knowledge graph automation is now better at handling real workspaces: folder selections expand correctly, graph status reflects when related maps need attention, and older notes can be surfaced for refresh when the graph engine has improved.

Consilience is also more careful about the facts it records. It is less likely to treat two names appearing near each other as proof of a relationship, and it flags questionable evidence instead of silently turning it into a confident connection. New workspace settings make these automation behaviors easier to review and control.

Cleaner entity pages

Entity pages now feel calmer and make better use of space. Sections can sit more naturally side by side without leaving large empty gaps, and common entity actions have moved into a single Actions menu in the header so the page stays focused on the content.

Layout controls have also been cleaned up with consistent, self-explanatory action cards, making it easier to understand what each option will do before applying it.

Clearer unfollow choices

Unfollowing a feed now gives you an explicit choice about what should happen to existing notes. You can keep them or remove them as part of the unfollow flow, with a clearer confirmation dialog that presents the choices in a simple order.

Consilience also now preserves your note-retention expectations after a feed is unfollowed, so notes are handled consistently instead of unexpectedly being kept forever.

More polished scrolling and overlays

Several small interface issues have been tightened up. Native scrollbars now remain fully visible in faded scroll areas, overlay controls no longer collide with nearby buttons, accent text is visible in the graph selection card, and unfollow confirmation now opens the proper modal.

Customizable entity pages

Entity profiles now have a cleaner, sectioned layout that you can personalize. Rearrange sections directly on the page, edit the real two-column structure in the layout dialog, and choose when to apply a layout beyond the current entity.

We also tightened the details: drag-and-drop feels more predictable, layout actions are easier to see, long sections scroll in place, and dialogs stay usable even with many results.

Live signals on the entity page

News, filings, market information, and other tracked sources now feel like part of the entity profile instead of a separate mode. The follow, filings, and retention chips act directly, and changes stay in sync across the entity page, Feeds panel, and graph cards.

Recent live items now use readable titles, and newly tracked entities start with a sensible retention setting that you can change at any time.

Source discovery that understands context

The source discovery flow has been consolidated into one prompt-first experience. It now starts with a clear suggested request and adapts its wording and suggestions to the kind of entity you are viewing, so events, organizations, and other entities get more relevant results.

Suggestions now include clearer localized reasons, better source coverage, and direct paths back to the related entity or Feeds view.

Cleaner company filing notes

Company filing follows now focus on material updates and produce more readable notes. Filing titles include more useful labels, note bodies preserve the important content and tables when available, and distracting noise has been reduced.

This makes filing feeds easier to scan while keeping vault growth under your control with retention settings.

Cleaner note analysis in the editor

Note analysis now lives in a compact status strip at the bottom of the editor, keeping outline, issues, key facts, related code, and word count close at hand without taking over the sidebar. This gives your notes more room while keeping contextual details easy to reach as you write and review.

Search has also moved into the title bar, and list rows are denser so you can scan more of your workspace at once with less scrolling.

More reliable Windows install and removal

Windows builds are now recognized more consistently as coming from Consilience, including during uninstall. This should reduce unnecessary warnings and make installing, updating, and removing the app feel smoother on Windows.

Ask your graph directly from the composer

Consilience now brings knowledge graph actions into the composer as slash commands. You can create entity briefs, timelines, overviews, tension reports, find related entities, connect ideas, see what changed, follow topics, watch updates, clip URLs, and ask for suggestions without leaving your current thread. Commands now handle @-mentions more naturally, show clear progress, render as proper action chips, and report honest empty or error states instead of silently doing nothing.

Entity pages also include quick launchers for relevant actions, so the next step for a person, company, concept, or topic is one click away.

Much smoother graph view for large workspaces

Large knowledge graphs open and rebuild more smoothly, with fewer stalls and no more blank canvas while Consilience is getting the view ready. The graph now keeps a clear “building” state until it is actually visible, restores settled views more reliably, and avoids saving a layout to the wrong workspace after switching.

Search, zooming, framing, and first-open behavior have all been tightened so big workspaces feel more responsive and predictable.

Cleaner launches, updates, and localized guidance

The app no longer flashes a temporary “Untitled” tab during startup, Windows users should no longer see command windows flicker when opening workspaces or files, and update prompts appear once per app session instead of once per window.

We also refreshed command guidance and product terminology across supported languages, including clearer chat messages for longer-running commands and consistent use of “workspace” throughout the app. The Academy and use-case pages have been rewritten and are easier to reach, making it simpler to learn what to do next.

Live feeds and on-demand web lookup

Consilience can now bring live sources into your workspace, so ongoing news, updates, and market-related signals can stay connected to the entities you care about. Feed activity is easier to scan with grouped, collapsible sections, clearer counts, and a dedicated place to add your own source.

Entity pages also now include a “Look up now” flow for fresh web results you can review and clip into your vault. Live folders are easier to recognize in the file tree, and price-style signals now support more asset types alongside existing tracked entities.

Exploratory scenarios in chat

You can now use /scenario to explore how a situation might unfold based on the material in your workspace. Scenario responses are framed as exploratory analysis, not predictions, and now lead with a prominent reminder before you open the full report.

Scenario results have also been strengthened with clearer grounding in your saved notes and more explicit checks around assumptions and uncertainty, helping you use them as thinking tools rather than forecasts.

Faster opening and clearer source trails

Large workspaces now open much more smoothly, with fewer chances for the app to appear stuck while your notes are being prepared. Several import and feed parsing issues were also fixed, so titles, dates, punctuation, and special characters are preserved more reliably.

Source trails are easier to follow: relation evidence can now take you directly to the exact sentence in the original note when available. We also fixed the Cases keyboard shortcut conflict, refreshed Academy content, and corrected a localization issue that could show Korean text in suggestion UI for users in other languages.

Consilience Academy

Consilience Academy is now available as a full self-paced course that teaches the app end to end, from core concepts to everyday workflows. Lessons use clearer language, practical examples, and live product previews so you can learn by seeing the real interface in action.

Academy is easier to find from the website navigation, and the chat lessons now better reflect how Consilience works: relationships and references are surfaced automatically, without asking you to manually create links.

App mentions in the composer

You can now mention installed apps directly from the composer when you want Consilience to focus a task on a specific application. The mention menu has been unified, making it easier to choose the right context whether you are referring to a workspace item or an app on your computer.

A more flexible properties popover

The properties popover can now be moved by dragging its header. This helps keep the content you are editing visible when the popover appears in the way, without interrupting your flow.

More reliable image generation

Image generation is more resilient to temporary failures, especially on Windows. If something briefly goes wrong, Consilience is better at recovering so your request is more likely to complete successfully.

Export your graph as an image or looping video

You can now save the graph view directly from View Settings as a crisp still image, animated GIF, or looping video. Transparent backgrounds are available for still images, making it easier to use your graph in slides, documents, and websites.

Animated exports now include controls for quality, frame rate, rotation speed, and motion style, so exported graph showcases look closer to what you see in the app and can be tuned for either polish or smaller files.

Follow revisions across notes

Consilience now does a better job of recognizing when a later note changes, replaces, or withdraws an earlier decision. The knowledge graph surfaces these relationships in a reviewable Revisions area, and entity pages now show a direct “Superseded by” link when a decision has been replaced.

It also reads dates from the body of a note, not just metadata or filenames, so older and newer decisions are ordered more accurately even in less structured vaults. This makes it easier to answer “what changed, and where?” without losing the original context.

More context for why something was decided

The knowledge graph now captures more of the meaning around statements: who made them, whether they express a decision, question, claim, quote, preference, value, or stance, and when one statement supports, challenges, or answers another.

This makes entity pages and AI answers better at explaining not just what you know, but why a conclusion was reached and whose perspective it came from. A new setting also lets you choose whether knowledge extraction should prioritize lower cost or higher quality.

Exclusions are immediate and reliable

Excluding a note from extraction now fully removes already-created knowledge for that note and refreshes the file tree right away. Excluded files are also consistently left out of search, suggestions, chat context, and interrupted-import prompts.

Ignore patterns are handled more reliably across platforms, and Windows paths in newly created wikilinks now resolve correctly. We also fixed a preview issue where an unfinished diagram or math block could hide the rest of a note until the closing fence was restored.

Case Files for investigations

Consilience now lets you create Case Files: persistent spaces for investigations that keep the question, relevant entities, and related conversations together. You can start a new case from the Cases view, pin a conversation into a case, or add an entity from its profile page so work no longer gets scattered across one-off chats.

Cases are easier to return to, too. Open the Cases view with ⌘⇧K, find cases by name from the command palette, and review what has changed since your last visit with a built-in investigation log.

Answers with better context over time

Your knowledge base now carries clearer time context. Consilience can show when an entity was first observed, understand the dates your notes are about, and answer questions about what is new or recently changed with a visible freshness cue.

It also does a better job preserving your own wording around decisions, open questions, claims, and quotes, making questions like “what did I decide about pricing?” more direct and easier to answer. Incorrect connections are easier to mark as wrong, with that flow now localized across supported languages.

Smoother design editing and export

Design work is more responsive and predictable. The canvas now uses clearer element names, offers a better picker when items overlap, keeps editing popovers within view, highlights changed elements after updates, and makes slide management easier from the filmstrip. Large decks should feel lighter while you work.

You can now export a single slide or page as a PNG directly from the canvas. We also tightened selection chips, restore prompts, inspector edits, and sidebar behavior so design conversations and edits feel less cluttered and more dependable.

Easier attachments, research, and writing

You can paste files directly into the composer, in addition to adding them by drag and drop. Attachment chips are cleaner in narrow sidebars, avoid duplicates when context is already selected, and are easier to remove in one step.

Research now presents its flow in your app language and writes results in the language of the topic when appropriate. Markdown table editing is smoother, caret movement is more reliable, and several editor and code-view details have been polished for day-to-day writing.

Mentions in conversations

You can now use @-mentions in user messages. Mentions appear as clear, distinct chips in the conversation, making it easier to reference people or items and keep context visible as a thread grows.

More reliable articles and changelogs

Articles and changelog posts now display diagrams inline when they are included, making explanations easier to follow without leaving the page.

We also fixed an issue where some newly published posts could fail to open. New content should now appear more reliably as soon as it is published.

Add knowledge from the composer

You can now bring documents, audio, and voice into Consilience directly from the composer. Instead of switching to a separate import flow, add source material as part of the conversation and Consilience will guide you through turning it into usable knowledge.

Chat is also more aware of ongoing graph builds. If you send a message while Consilience is still processing new material, it can hold the message and send it once the graph is ready, so answers are less likely to be based on partial context.

Smarter mentions and clickable references

@-mentions are easier to use across the composer. The picker now surfaces better matches for files and entities, keeps the keyboard selection in view, and makes long names readable with compact chips. Mention chips now show clearer type icons, can be opened with a click, and are easier to remove or delete cleanly.

Replies are more connected to your workspace too. When Consilience refers to a file it created or found, that reference is now easier to open directly from the conversation.

New ways to browse entities

The knowledge graph now includes a dedicated entity table for exploring everything Consilience has found in a workspace. Open it from the launcher or title bar, search by name, sort the list, and jump from any row into a full entity page with its surrounding context.

Entity pages and graph details are more consistent, and renamed documents now keep their displayed titles in sync. Consilience also does a better job answering precise questions about the graph and showing relevant supporting context when entities are mentioned.

A richer workspace launcher

The welcome screen now presents recent workspaces as visual cards with miniature knowledge-graph previews, workspace stats, and quick access to common actions. Larger workspaces are easier to tell apart at a glance, and the launcher feels more like a dashboard for returning to active work.

This release also includes everyday polish across the app, including cleaner copied paths on Windows, clearer setup prompts when an optional local tool is missing, better map pin sizing, and more reliable graph loading when opening entity views directly.

Verifiable graph evidence

Relationships in the knowledge graph are now easier to trust. Hover over a relation to see the sentence in your note that supports it, and graph-generated context now cites source notes by title so you can jump back to where a claim came from.

Consilience also does a better job reading everyday Markdown. Links, emphasis, code spans, tables, imported-note banners, and front matter are handled more cleanly, which helps keep entity names and relationships tidy.

Entity types and saved collections

A new Entity types tuning area lets you review how your graph is organized, find entities whose type is still unknown, and assign or adjust types directly from entity chips and graph cards. The interface now makes editable values clearer and keeps the same card-based design language as the rest of the app.

You can also build saved collections from the graph: start with a kind of entity, narrow it around an anchor, pivot across related entities, and save the resulting set for reuse.

Faster search and a catch-up digest

Search should feel more responsive, especially after editing or importing notes. Results are available sooner while Consilience keeps them up to date as your knowledge base changes.

This release also adds a “since you were last here” digest that highlights newly discovered entities, changed notes, and notable shifts in your knowledge graph so it’s easier to re-enter a growing workspace.

More dependable sharing, checkout, and captures

Sharing is more reliable: Consilience no longer folds previous share bundles back into new exports, and it handles oversized share links more gracefully. Returning to the app after checkout is also more consistent.

App captures are improved across macOS and Windows, with more complete page content, better shortcut recording, and clearer prompts when the app needs permission to capture what you asked for.

Suggestions that meet you where you are

New chats now offer concrete, workspace-aware ideas directly above the composer, helping you start with something useful instead of a blank page. Tapping a suggestion starts the work for you, and suggestions now share the same clean surface as composer actions so the interface feels calmer and more consistent.

Suggestion activity is also easier to control. Consilience is more conservative about background suggestions, and you can turn them off when you want tighter control over usage.

Clearer graph answers and entity cards

Consilience is better at answering precise questions about your workspace, such as how many notes mention something, what is connected to an entity, or how two items relate. Results are presented as direct answers rather than vague matches, and fast completed steps now show more precise timing so progress feels more transparent.

Entity cards have also been redesigned to be worth opening. They highlight the most useful relationships first, show where information came from, and sit more calmly alongside the chat composer.

Workspaces in Search Everything

The global search palette now includes workspaces. Selecting one lets you choose whether to switch the current window or open it separately, with the usual save prompt when needed, so moving between projects is faster and safer.

More room on the Free plan

We increased the Free plan allowance so larger imports and longer workspace sessions are less likely to stop midway. When you do reach a limit, the app now explains what happened clearly and offers the next best option, including starting a Plus trial when available.

GitHub actions in Source Control

The Source Control panel now includes a guided flow for connecting to GitHub, publishing a project, and opening a pull request when your workspace is ready. Status checks are clearer and less noisy, so you can see what action is available without leaving Consilience.

Staged changes in diffs

You can now review staged changes directly in the diff viewer before committing. This makes it easier to verify exactly what will be included in your next commit, especially when you are splitting work across multiple commits.

More reliable maps

Maps are now more stable when switching themes, opening the large map view, or returning to a conversation with an inline map. This release fixes cases where maps could freeze, appear blank, or interfere with each other after changing views.

Stability and interface polish

Consilience now handles several edge cases more gracefully, including stalled background work, unavailable workspace locations, oversized previews, and sign-in interruptions. Notifications and dialogs have also been made more consistent across the app, giving you clearer feedback when an action succeeds, fails, or needs your attention.

Clearer credit purchases

Buying usage credits is now easier to understand before you check out. The purchase screen supports preset bundles as well as custom amounts, and eligible volume discounts are shown up front so you can see the final price before continuing.

We also fixed an issue that could cause the browser-based checkout flow to get stuck, making credit purchases more reliable from start to finish.

More reliable Windows titlebar buttons

On Windows, the custom titlebar now correctly responds when you click directly on button icons. Minimize, maximize, and close actions should feel consistent no matter where you click within each control.

Usage credits for uninterrupted work

You can now add prepaid usage credits to keep working after your plan’s included usage is reached. Instead of being stopped mid-flow, Consilience can continue serving requests and draw from your credit balance when you’ve chosen to enable it.

This makes heavier research sessions smoother and gives teams more flexibility without needing to change plans immediately.

More accurate automatic tagging

Automatic tagging and knowledge-graph extraction have been improved, with more relevant results and less cleanup required after importing or analyzing content.

Usage accounting for these actions is also more consistent, so credit consumption should feel clearer and more predictable.

Expanded language support

Consilience now includes broader multilingual support across the product, including account pages, changelogs, and marketing pages. Spanish, French, German, Japanese, and Chinese are now available in more places.

A new language menu in the navigation bar makes it easier to switch languages, and localized pages now render more reliably in the selected language.