Changelog

Most of this release is invisible - but this time it's about something more immediate than infrastructure. We went looking for every place where Conception could lose your work, and closed them.

Nothing here adds a button. What it adds is confidence: that the pane you opened stays open, that the sentence you typed a second before closing the tab is still there when you come back, and that a note you moved into a folder stays in that folder.

Panes That Stay Put

The sliding columns are the core of how Conception works, so we rebuilt the model underneath them. Panes now have their own stable identity rather than being tied to whatever URL they happen to be showing, which fixes a family of bugs at the root instead of one at a time.

• Asking your first question no longer breaks the pane. Typing a question on a fresh chat and hitting enter could leave you staring at a 404 page while the answer generated invisibly behind it. That's fixed - the answer streams in where you asked it.
• Exiting full screen keeps your pane. Expanding a pane and collapsing it again used to drop you back with the pane gone. It now carries its expanded state through.
• Two tabs no longer fight. Panes are now tracked per browser tab, so opening a second tab doesn't overwrite the layout in your first. Your last layout still seeds a brand-new tab, then the two go their own way.
• Each pane scrolls on its own. Long documents scroll inside their own column instead of moving the whole row, and a sideways scroll still works even when your cursor is over a pane.

Saves That Actually Land

Conception saves in the background as you write. Two gaps in that mechanism could quietly cost you work, and both are now closed.

Closing a tab now saves instead of discarding. Both note edits and structure changes - moving a note into a folder, reordering your sidebar - were held briefly before being sent to the server. Close the tab inside that window and the change was gone. Hiding a tab or closing it now flushes whatever is pending, on its way out.

Your links stop disappearing. Opening a note on a second device could trigger the app to send an empty copy of that note's internal links back to the server, wiping the connections between your pages and leaving stray entries in its version history. The cycle is closed, and notes now load with their real links and text from the start.

Failures are visible now. When a structure change can't reach the server, Conception retries, and then tells you - with a banner and a retry button - instead of showing a spinner that finishes as if everything saved. If your files can't be loaded at all, you get an error and a way to try again rather than an empty sidebar.

New accounts land where they should. A routing bug meant every new user was sent to the chat page instead of the Getting Started note written for them. Fixed.

Version History, Corrected

Version history had a set of quiet inaccuracies, all of which we've fixed:

• Simply opening the history dialog used to create a version, stamped with the name of whoever opened it and bumping the note's version number. Looking at history no longer changes it.
• Restore is disabled while a save is in flight, so a restore can't be immediately overwritten by an autosave that was already on its way.
• Plan-based access is now judged by a version's age, and per version - previously the first entry on every page of history appeared unrestricted, whichever page you were on.
• When a restore isn't permitted on your plan, the message now says so instead of failing silently.

Databases That Stay in Their Lane

Editing a row is now strictly scoped to the database it belongs to. Previously a malformed or misrouted request could write over the contents of another note you own, and deleting a view could - through a miscounted safety check - remove the only view of a different database. Both paths are closed, and an operation that doesn't match anything now returns a clear error rather than reporting success it never performed.

Database rows also behave properly in search: opening a row from the command palette now takes you to the row, rather than to a broken chat page, and it shows the right icon while you're choosing. Optimistic row edits reconcile with the server correctly, and bulk operations have a sane upper limit.

A New Search Engine Underneath

Search in Conception has, until now, worked by downloading a copy of your whole workspace into your browser and searching it there. That was fine at a few dozen notes. It isn't at a few thousand.

We've replaced the engine. Search now runs in the database, over a single index that covers note titles, tags, page properties, note bodies and chat messages together - with fuzzy matching, so a near-miss or a typo still finds the page. It's built, tested, and live behind the scenes; the command palette switches over to it in the next release, and you should feel the difference most if you have a large workspace.

In the meantime, the palette itself got safer and steadier. Search results are now rendered as text under all circumstances, closing a flaw where a page containing markup - including content pasted in from the web by an AI - could execute it in the palette of anyone who searched for that page. Typing characters like ( no longer throws an error mid-search, and the search mode is now judged fresh on every keystroke rather than lagging a character behind.

One New Model

Kimi K3 from Moonshot AI joins the lineup, following the K2.6 addition in our last update. As always, it went through the evaluation suite before reaching the model picker.

Voice Input

You can now talk to Conception. Hit the microphone in the chat input and start speaking - a live waveform shows you that we're listening, and when you stop, your words are transcribed straight into the input where you can edit them before sending.

It's built for the way ideas actually arrive. Dictate a half-formed thought, keep typing to sharpen it, then send. Transcripts append to whatever is already in the box rather than replacing it, so you can layer voice and typing together. Changed your mind mid-sentence? Discard the recording and nothing is sent.

A Refreshed Model Lineup

We rebuilt the model picker from the ground up this cycle. The headline: several new frontier models, two new providers, and a long list of older models retired.

New this cycle:

• xAI Grok 4.5 - xAI's most intelligent model, built for coding and agentic work.
• OpenAI GPT-5.6, in three variants tuned for different jobs:
  • Sol - for demanding reasoning and long-horizon work.
  • Terra - frontier intelligence balanced against cost.
  • Luna - the fastest of the three, for cost-sensitive everyday work.
• Anthropic Claude Fable 5 - Anthropic's flagship model for the hardest problems.

On Fable 5 - in our last update we had to remove it shortly after launch, when a U.S. export control directive required Anthropic to suspend access for foreign nationals. We said we'd restore it the moment it became available again. It's back, and it's available to Pro and Premium users today.

We also welcomed two new model providers to Conception: Z.ai, with GLM-5.2, and Moonshot AI, with Kimi K2.6 - both strong at long-horizon reasoning and coding, and both a meaningful widening of the range of choice available to you. They join DeepSeek V4 Pro, Claude Sonnet 5, Claude Opus 4.8, Gemini 3.1 Pro, and the rest of the current lineup.

Just as importantly, we retired a long tail of older models - the GPT-4 series, Claude 3 and Claude 4 generations, Grok 3, Gemini 2.5, and others. Keeping them around padded the list without helping anyone choose. Every model in the picker now is one we'd actually recommend for something.

The picker itself is clearer, too. Each model carries its provider's icon, newly added models are badged, and anything outside your plan is marked up front rather than failing after you send. If you'd rather not think about any of this, Best stays at the top of the list and picks a model for each query on your behalf.

Attachments That Follow the Model

Images and PDFs now work properly in chat and search. Attach a screenshot, a scanned page, or a research paper, and models that can see them will read them directly as part of your question.

The file picker also adapts to the model you've selected. Not every model can read images, and not every model can read PDFs - so instead of letting you attach a file that would quietly fail, Conception now only offers the formats your chosen model can actually handle. We've also broadened the set of supported image formats along the way.

Guardrails: How We Test the AI

Adding a model to Conception used to be a judgement call. It isn't anymore.

Every model now has to pass an automated evaluation suite before it reaches you, and again whenever its provider updates it. We measure four things:

• Quality - each model is graded against a fixed set of real tasks, so a provider's silent update can't quietly make your results worse without us noticing.
• Safety - we run adversarial tests to confirm that a malicious webpage, a pasted document, or a crafted note can't hijack the assistant into ignoring its instructions or leaking your data.
• Speed - we benchmark how quickly each model starts responding, because a brilliant answer that takes thirty seconds is often the wrong trade.
• Cost - so we can keep capable models on the free tier and keep them there sustainably.

The results decide which models we ship, which model powers which part of the product, and which ones we drop. Several of the retirements above came directly out of this process - models that no longer earned their place once we had the numbers in front of us.

This is the same instinct behind last cycle's security work: the parts of Conception you can't see should be the parts you can most rely on. We'd rather tell you how we test the AI than ask you to take our word for it.

Security Foundations

This release is different from our usual updates - almost everything in it is invisible. Over the past few weeks we've rebuilt the security foundations underneath Conception, strengthening how your data is authenticated, isolated, and tracked. You won't see new buttons, but everything you create is now protected by multiple independent layers.

Stronger authentication. Every request between the app and our API is now verified using cryptographically signed tokens issued when you sign in. Identity is checked end-to-end on every call, so the API always knows - and verifies - exactly who is asking before touching any data.

Isolation enforced at the database layer. Your notes, chats, and databases are now protected by row-level security directly inside the database itself, in addition to checks in the application. Every read and write is scoped to your organization at the lowest layer of the stack, so even in the unlikely event of an application bug, data can never cross from one tenant to another.

Introducing organizations. Under the hood, workspaces now live inside organizations - a new top-level boundary that owns every piece of data you create. Today this happens automatically and changes nothing about how you work, but it lays the groundwork for what's coming: teams, member roles, and shared workspaces, all built on a foundation where access control is explicit from day one.

A complete audit trail. Every meaningful change - creating, editing, or deleting notes, chats, databases, and account settings - is now recorded in an append-only audit log that even we can't modify. This is the kind of accountability infrastructure that larger teams and compliance frameworks (like SOC 2) expect, and we're building it in early.

We treat the security of your ideas as a feature, not a checkbox. More of this foundational work is on the way.

AI Model Updates

We've continued expanding the range of AI models available in Conception. Claude Opus 4.8 and xAI's Grok 4.3 are now available for Pro and Premium users, giving you access to stronger reasoning, improved writing quality, and a broader set of tools tailored to different workflows and use cases. Our goal remains to provide flexibility and choice, so you can select the model that best fits the task at hand.

Earlier this month, we also introduced Anthropic's Claude Fable 5 before subsequently removing it for affected users due to a new U.S. export control directive requiring Anthropic to suspend access to Fable 5 and Mythos 5 for foreign nationals. While this interruption was outside our control, we remain committed to transparency around the models available in Conception - including how they handle data and any limitations that may apply - and we'll restore access to Fable 5 immediately should it become available again.

Powerful Databases, Seamlessly Built In

Databases have officially arrived in Conception - bringing powerful structured workflows into the same workspace where you already write, research, and think. You can now organize information with flexible tables, custom properties, filtered views, and scalable systems built for projects, pipelines, trackers, knowledge hubs, and more.

Databases combine the simplicity of notes with the power of structured data, making it easier to manage everything from tasks and content calendars to CRM workflows and research repositories. Create rows as records, customize fields to match your workflow, sort and filter information instantly, and turn growing collections of data into clean, usable systems.

Built natively into Conception, databases work alongside pages, notes, chats, and folders - so your structured work and unstructured thinking stay connected in one place. No context switching, no separate tools, no fragmented workflows.

This launch marks a major step toward Conception becoming the complete workspace for organizing curiosity.

Unlimited Notes for All Users

We’ve removed note limits for both Free and Pro plans. All users can now create unlimited notes across their workspace.

Whether you use Conception for quick capture, long-term research, journaling, or managing complex projects, you no longer need to think about note caps. Create freely and scale your workspace without restrictions.

Workspace Navigation Redesign

We’ve significantly redesigned navigation across the app to create a faster, cleaner workspace experience.

Chat history now lives directly inside the left sidebar, while the sidebar itself has been reorganized to separate File Explorer and Chat History into dedicated tabs beneath Quick Access actions. This keeps navigation clearer as workspaces grow and makes switching between content types much faster.

The Search action has also been moved from the Quick Access panel into the top navigation bar for easier global access.

Cleaner Editor Experience

The right sidebar has been removed entirely to create a more focused writing and browsing experience with less visual clutter.

Table of Contents has been rebuilt as a floating editor plugin, giving users quicker access while preserving screen space. Table of Contents is now also supported inside both notes and chats, making long-form content easier to navigate wherever you work.

New AI Models

Expanded model selection with two new additions:

• Added ChatGPT 5.5
• Added Claude Opus 4.7

These models are now available in supported AI workflows across chat and content generation.

Website Updates

Updated the public website to reflect the latest product direction and feature set by adding a dedicated Databases product page that highlights structured workflows, updating pricing and app messaging to reflect unlimited notes, and refreshing the homepage to showcase new database functionality and expanded workspace capabilities.

Note Export

You can now export notes in multiple formats, including Markdown & CSV, HTML, and PDF. This makes it easier to share your work, create backups, or move content across different tools without losing structure.

The export system also gives you more control over what gets included. You can choose to export the entire page or exclude files and images depending on your needs. In addition, notes that contain linked subpages using [[...]] are now recognized during export. You can choose to include these subnotes, and if enabled, Conception can automatically organize them into folders, preserving the structure of your content outside the app.

Folder Page Improvements

We’ve made several improvements to how folders behave, especially when it comes to organizing and interacting with content.

Rearranging items inside a folder is now much more intuitive. You can drag and drop notes, chats, and subfolders directly within the folder page, similar to how it works in the sidebar. Any changes you make are instantly reflected across the workspace, ensuring that the folder view and sidebar stay in sync.

We’ve also fixed an issue in gallery view where dragging an item would accidentally select its internal content. Interactions are now more precise, so dragging behaves as expected without interfering with the content inside notes or chats.

Another important improvement is the introduction of a consistent structure within folders. Content is now always organized in a defined order, with folders appearing first, followed by notes, and then chats. When new items are created, they automatically follow this structure instead of being appended arbitrarily. This resolves earlier inconsistencies where new folders or notes would appear in unexpected positions.

Cover Image Fix

We’ve resolved an issue where updating cover images was failing in some cases. Cover images should now update reliably across notes.

Behind the Scenes

This update also includes several internal improvements to ordering logic, drag-and-drop behavior, and synchronization between different parts of the interface. These changes lay the groundwork for a more consistent and scalable workspace experience as new features are introduced.

This update introduces a more structured way to organize and standardize your workspace - with templates and folder-level intelligence at the core.

Page Templates

You can now create reusable templates for your pages.

• Create structured starting points for notes
• Save time on repetitive setups
• Maintain consistency across your workspace

Templates can include:

• Page properties
• Pre-filled content
• Formatting and structure

Folder Templates

Folders are now smarter.

Each folder can define its own default template for notes created inside it.

• Any new note inside a folder automatically follows the folder template
• Enforces consistent structure at the folder level
• Great for workflows, projects, and documentation systems

Folder Pages (New)

Folders now open as full pages.

• New route: /folder/:id
• Clicking a folder opens a dedicated view
• Displays:
  • Notes
  • Chats
  • Subfolders

Views available:

• List view
• Icon view
• Gallery view (with previews of notes & chats)

Folder-Level Defaults & Enforcement

Folders can now enforce:

• Page properties
• Initial content (via templates)

When:

• Creating a new note inside a folder → template is applied automatically
• Dragging a note into a folder → user is prompted before:
  • Overwriting existing properties
  • Injecting template content

Template Management System

We’ve laid the foundation for a template ecosystem:

• Templates are now stored at workspace level
• Dedicated schema for templates
• Supports versioning, metadata, and future extensibility

Includes:

• Public/private templates
• Version tracking
• Structured content storage

Also introduced:

• Workspace ↔ Template relationship layer (preparing for future template marketplace)

Page Properties (Front Matter)

Pages in Conception can now contain structured metadata through Page Properties (front matter). This introduces a flexible way to add typed fields directly to pages, allowing notes to evolve from simple documents into structured knowledge entries.

Properties appear directly below the page title and can store different data types such as text, numbers, checkboxes, URLs and many more. These fields remain fully editable and are stored as part of the page metadata, making them ideal for organizing research notes, tracking projects, or creating lightweight databases inside your workspace.

Folder-level property templates can also be defined to enforce consistent structure across pages within a directory. When a template is applied, newly created pages automatically inherit the predefined property schema, ensuring uniform metadata across related notes.

This feature enables users to introduce structure without sacrificing the flexibility of freeform pages. Over time, workspaces can evolve from collections of notes into interconnected systems of structured knowledge.

New AI Models Available

Added support for several new AI models to expand reasoning, speed, and generation capabilities across chats and AI-assisted workflows.

The following models are now available:

• GPT-5.4
• Gemini Pro 3.1
• Claude Sonnet 4.6

These models improve reasoning quality, context handling, and response performance across research, writing, and knowledge exploration tasks.

Supported models can be selected directly from the AI model selector when starting a chat or running AI commands.

Website Performance Improvements

Improved the performance of the Conception website by optimizing large image assets across landing pages and documentation.

Previously heavy images were replaced or compressed with lighter assets to reduce load time and improve rendering speed across devices. This change significantly decreases page weight, allowing faster initial page loads and smoother navigation, particularly on slower networks.

These optimizations improve the overall user experience while maintaining visual clarity across marketing and product documentation pages.

Notes Version History (Version Control)

You can now view, compare, and restore past versions of your notes. Every meaningful change is automatically saved in the background, so you can rewind edits, preview differences, and recover previous content with confidence - no fear of overwriting work or losing ideas.

Version history now runs automatically in the background, creating intelligent snapshots while avoiding duplicate saves through content deduplication. Changes are presented in a chronological timeline with clear timestamps and summaries, making it easy to understand how a note evolved over time. You can preview any version, compare it against the current state with inline diffs, and restore it when needed. Manual save points are supported alongside debounced auto-saves, ensuring important edits are always captured without adding friction.

Directory Version Control

Folders now have history too. Structural changes like renames, moves, and re-organizations are tracked, giving you visibility into how your workspace evolves over time. If something goes wrong, you can trace changes back and restore with clarity - perfect for large workspaces and long-running projects.

Directory versioning introduces visibility into structural changes across your workspace by tracking directory-level mutations over time. This makes large refactors and reorganizations safer by allowing changes to be reviewed and traced if needed. As workspaces grow in size and complexity, directory history improves auditability and reduces the risk associated with structural updates.

Trash Bin for Notes & Chats

Introduced a Trash Bin system to safely handle deleted notes and chats, enabling recovery and permanent deletion workflows. Deleted items are no longer immediately removed. Instead, they are moved to Trash, where users can review, preview, restore, or permanently delete them.

A dedicated Trash entry has been added to the sidebar dropdown to centralize access to deleted content. Notes and chats are now soft-deleted and moved to Trash instead of being immediately removed. The Trash dialog lists deleted items in reverse chronological order and provides read-only previews—notes render in a non-editable state, while chats display only historical questions and responses with all interaction controls disabled. Items can be restored to their original workspace path and hierarchy, or permanently deleted for irreversible removal. Access controls are enforced to ensure users can only view and manage their own deleted content.

Added Opus 4.6 model

Integrated the Opus 4.6 model into the AI model selection, making it available for supported chat and content-generation workflows.

We’ve launched Incognito Mode to give you a more private way to search and chat, making it easier to work on sensitive ideas or explore freely without leaving a trace. You can quickly enable or disable this mode anytime using the shortcut CTRL + SHIFT + I, ensuring seamless control over your privacy.

Alongside this, we’ve significantly expanded our AI model offerings by adding powerful new models from OpenAI, Anthropic, xAI, and Gemini, giving you more choice, better performance, and improved results across a wide range of use cases.

We’ve also fixed important issues related to refund cancellations and recurring payments, and enabled 3D Secure for all card transactions, adding an extra layer of protection and reliability to your billing experience.

Resolved an issue where note's content were being duplicated after page refresh or when reopening slides, caused by stale cache data and multiple save events triggering simultaneously. The fix optimizes the note-saving flow by ensuring each save is processed only once, with improved cache validation and synchronization between client and server. This update prevents content corruption and ensures note data remains consistent and reliable across refreshes and slide interactions.