Changelog

Piece text now renders in a cleaner, more legible typeface with better contrast, and lists sit snugly under their items instead of trailing extra space — while staying clearly spaced from the paragraphs around them.

Each stage of a master's timeline now names what you'll actually do — and every button tells you where it takes you before you click.

The option to add presentation slides now lives quietly in the Introduction header, appearing only when you need it, instead of taking up its own row on the page.

Opening a French playbook no longer greets you in English. The AI's opening line now follows the language of the playbook, master, or piece you're actually in.

A stalled response could leave your coach spinning indefinitely and, on reload, show your message twice. Both are fixed — a response now always completes or clearly errors out, and only one reply appears.

The first time you connect, aïno greets you before dropping you into the workspace — a short, personal welcome tailored to whether you're learning or building, unfolding at a human pace instead of arriving all at once. Pick the conversation back up later and it won't repeat a greeting you've already had.

You can now reset your conversation with your learn coach, the same way you can in the editor — your progress and history are kept, only the transcript clears.

The learning view now has warmth and depth instead of bare margins — a quieter, more deliberate frame around your work.

If you're learning inside an organization, your home view now centers on your coach rather than a design workspace built for editors.

Your path through a master now shows as a staged timeline instead of a breadcrumb, with a floating coach ready to step in for a conversation without losing your place. Moving between your journey and that conversation flows as one motion, and closing it is a single tap or an Escape key away.

Coaching now draws only on the playbooks you actually hold a seat in, and adjusts to your real training status and focus. At the organization level, it opens with an honest introduction instead of assuming you're already underway.

On the learning page, the conversation now sits right against your message box instead of leaving a gap above. Open a master directly and the coach bar is there too, ready with the right conversation for it.

What you're working toward now stands out up front, the next-step button sits in a consistent place, and labels and actions across your master's home are sharper — including a clearer name for what's left to practice.

Start a review with a clean, focused space — paste an email, drop a file, or write directly, and the piece editor opens instantly, cursor ready, with every keystroke and paste registering properly. Until you bring something, you're guided on what to do and where to click; once you do, your real piece and the two-pane view appear.

If a piece's original conversation had been removed, opening "Your pieces" could fail entirely. Those pieces are now quietly left out instead of taking down the whole list.

Your progress timeline no longer overlaps the status marker in the corner — each stays legible on its own.

The first-run reviews screen is now properly centered, with a tighter, less repetitive introduction to the practice, and its empty state reads in full instead of cutting off mid-message. Deleting a review also now works every time.

Confirmation and error messages now follow your theme consistently, with a clear color cue for success, warning, or error. French messages also read correctly, with the right article every time.

Live simulation runs now look and feel like an actual phone call, from the pre-call brief through a live call stage to your debrief — which now flags when your own confidence didn't match your score.

If you own a playbook, you can now invite people onto it directly — no need to go through an admin. Admins keep full reach across the organization, and a failed invite preview now tells you what went wrong instead of leaving you stuck.

Opening your organization now shows the right workspace immediately, with no flash of the wrong layout first. Links your coach offers to a master or playbook are cleaner and only ever appear when they actually lead somewhere.

Comment on a real piece, bring your own standards for your coach to work from, or just start talking — all three now stand on equal footing instead of pushing you toward one, and the greeting lays them out one per line so you can actually see your options.

New pieces in your personal space now start in your browser's language instead of always defaulting to English.

The learner home no longer shows a stray, empty toolbar, and its pages now share a more consistent look.

The sign-in screen now speaks to craft and pride in the work, not generic AI content talk — and the tagline reads unambiguously as aïno's promise to your team, not aïno praising its own work.

New tables start clean, with no header row forced on you — add one only when you want it, and switch it back whenever you like. Turning a row into a header now reveals a blank row beneath it instead of overwriting what you'd already written.

Hover over a pasted image to expand it full-screen or remove it, without leaving the editor. Images in your reviews, training pieces, and how-tos also stay put on their own — no more disappearing on you.

Code blocks now look and behave consistently across chat, your pieces, and read-only views — long lines wrap cleanly, and a copy button is always at hand.

Cancelling or deleting a conversation you don't own is now correctly refused, instead of quietly appearing to succeed.

Your coach now writes in a more formal, high-register French — proper inversion, French quotation marks, and none of the casual tells that broke the tone before.

Your coach now draws on its full toolkit consistently — including looking up masters — without ever producing work for you, and it no longer claims to have updated a document it never touched. Its coaching moves are sharper too: acknowledging what's working, reframing setbacks, and naming mistakes plainly, without slipping into empty praise.

Skip the manual drawing — point at your reference and generation places labels and connectors for you, anchored to your key points, as a quick-start alongside manual upload. Arrows now land precisely on the key point they annotate, labels no longer get cropped off the image, and the result shows only the key points you kept, worded exactly as you wrote them. Once you've made your first one, Regenerate tucks into a small icon beside upload and delete, out of the way until you need it.

Design guidance now reasons about how your master's pieces fit together as one instrument, and can propose a reframe when that's the right move — not just flag what's missing.

Switching into learner view now shows a spinner while the page loads instead of sitting still, so you know your click landed.

The Delete action on a master's review-steps section was silently doing nothing. It now removes the section as expected, with the same optimistic feel as everywhere else.

Key points, steps, common errors, reframes, and storyboard slides now update live and in order — no more waiting on a manual refresh to see changes appear.

Live-simulation setup is now a clean, click-to-edit document instead of a stack of open editors — click any field, including grading criteria, to edit it in place. Deleting a training example now stays scoped to that example, so confirming one delete never dismisses another out from under you.

The simulation's microphone check was failing outright, leaving Start permanently disabled behind a confusing error. Mic access now works, and if it still can't reach your microphone, you get a clear reason why.

A search variant could break a conversation partway through a multi-turn exchange. Search now runs on a stable path that holds up through the full session.

Table headers across the editor are steadier to work with, with cleaner spacing and better keyboard support. In the users matrix, playbook labels now sit directly above their matching rings instead of drifting across the table.

Web search results now appear before the action built from them, so you see the supporting research first and the recommendation after.

Conversations are less likely to interrupt themselves, long ones no longer fail mid-stream on replay, and a rare double-send from hitting cmd-enter and clicking at once is gone — only one response ever runs. Cancelling a response or starting a new chat now completes cleanly every time, even if the conversation is already gone.

The editor for a sent message no longer squeezes into a narrow, off-center box — it now fills the width of the panel while you edit.

A gap in your coach's identity setup meant its connoisseur's attention to craft wasn't reliably reaching every learn session. It's back in place.

Hover a table to reveal row and column grips — click one to insert, move, or delete a whole row or column in place, with your table staying valid throughout. Handles now follow your editor theme so they read clearly in light mode, line breaks and bold or italic text inside cells stay put after you save and reload, and even large tables no longer stutter or drop keystrokes as you type.

Before your first message or photo is sent to the AI in Coach or Training, you're asked to confirm it — once. You can withdraw that permission any time from the space-switcher.

Paste or drop an image and it shows up immediately instead of after a delay — no more staring at a blank spot while it uploads. Oversized screenshots are now capped in height so they don't dominate the page.

Opening a piece no longer triggers a save of its own — a save now only fires once you've actually made an edit.

In Safari, clicking from the piece editor into the review box could leave the caret stuck in the piece instead of moving to where you clicked. Fixed.

Covers were failing to generate for a stretch. Image sizing requests are now sent in the format the model expects, so generation works again.

The Practice section now ranks your work into one column: your next set stands out, what's left dims back, and cleared sets recede quietly out of the way. Starting a session now shows a simple spinner in place of the button, instead of a distorted button that lost its label.

Previewing a piece now lets you copy or export it on the spot — no need to open the full piece first.

Bold text and headings in AI action cards now render as formatting instead of raw markdown symbols, and a long card tucks itself behind a "Show more" so it no longer takes over the sidebar.

Exporting a master's bundle now brings along its learner introduction — narration text and slide images included.

Reference, template, and intent fields now accept pasted or dropped images, just like the intro field always has.

The piece and your review now sit side by side, rendered in the same polished style the author sees while writing — so you study the real thing, not a plain fallback, and never lose your place scrolling back and forth. A stray table toolbar that used to sit over the review box and block your clicks is gone too.

Starting a blank master no longer opens onto an empty form. Three clear paths — bring your material, see real examples, or talk it through — lead straight into a conversation with your AI design partner.

Adding presentation slides to a master is now a single, clean link with no placeholder clutter, and once slides exist the prompt matches your master's state — jump straight into editing your deck.

While your deck generates, each slide visibly shimmers in turn so you can see exactly which one is being drawn, and a cue points you to the chat where the narration of what's being built streams in live. The deeper, two-minute planning pass shows the same clear progress on both canvas and chat — no frozen screens, no accidental double-starts.

Chat, craft, master, and learning conversations now load through one consistent path, reducing the chance of mismatched or stuck states.

You can now add any optional section straight from the header, at any point while writing — no need to scroll to the bottom of a long master first.

If your connection drops mid-render, the deck no longer sits blank waiting for a reload — it picks the slides back up on its own, and your place is kept even across a reload. A rejected illustration during redraw now gets the proper response instead of a generic error.

Answer a practice case and get your verdict and coaching together immediately, instead of a follow-up question — and the answer box now always appears for a new case, so you're never left facing a blank screen.

Recognition and understanding exercises now sit in one centered column with the piece front and center, feedback and the advance banner running the full width, and the verdict or question sitting directly beneath the example itself. Your grade streams in with a quiet loading indicator, scrolls into view the moment it lands, and stays on screen alongside your answer — with banners colored to match the verdict throughout.

Jump straight into a specific example when you begin a training session, instead of starting from wherever the arc puts you.

Back always takes you up to where you came from, and full-screen sessions close with a clear X — no more guessing which arrow does what.

The example list marks recognition cases with a green or red check and understanding cases with a lightbulb, grouped and counted correctly — and the same icons now appear on the format and verdict selectors, so your choice is unmistakable at a glance.

Ask curate for new training examples and it now suggests open-question understanding exercises alongside recognition ones, reading your collection's current mix and leaning toward whichever format it's missing.

Toggle an example between recognition and understanding and the right fields follow automatically — nothing left over from the old format to trip up saving.

Collection headers now use the same settings menu as playbooks and masters, with the focus line updating live as you edit. Collection detail opens as a two-pane view — an index of your examples on the left, full editing on the right — and opening a collection no longer flashes an empty state before your examples appear.

Recognition cases now always ask for a verdict, even when a leftover authoring question lingers, and AI-proposed understanding cases show their open question under a neutral badge instead of appearing mislabeled.

"Add master" now sits as a clear action in the toolbar and at the top of both the cover and settings menus, instead of a small floating button easy to miss.

Org admins could invite a learner to a playbook but couldn't cancel, resend, or review that invitation afterward. That gap is closed — org admins now have full control over invitations on their org's playbooks.

Inviting someone to a playbook now only offers roles you can actually grant, so an invitation never fails partway through — and new teammates arrive with the correct access from the start. Want to hand someone Owner? Invite them first, then promote them once they've joined.

The OK/Not OK tally is gone from the collection header since each row already shows its own verdict, and the OK / Not OK buttons now sit evenly matched, side by side.

The People matrix now scrolls past the fold, so rows that were previously cut off are reachable.

Opening a skill conversation no longer fails with an error — the greeting now loads correctly every time.

Draft your next message in the AI sidebar while a response is still streaming in, instead of waiting for it to finish.

Generated storyboards name the concrete piece you'll build and tie it back to the problem you started with, in a plainer voice with the robotic em-dash tic dialed back.

While your AI is working, the status text now shimmers along with the icon — a clearer sign that something is actively happening, not stuck.

Apply and Discard now outline and fill on hover, just like Accept all, so it's unmistakable which action you're about to take.

Cover banners are now generated natively at the right shape instead of being cropped down from a wider image, so nothing gets cut off. The description field now holds up to 1500 characters, with a counter as you approach the limit, so you can brief a cover properly and submit without surprise errors.

Copy a table from a document or spreadsheet and paste it into your piece — it now arrives as a proper table attachment instead of broken formatting. Short lists and headings still paste inline, as expected.

Your coach's replies come back faster, and files you attach stay available even if a link would otherwise expire.

When your coach proposes a presentation from a storyboard, it now appears right where you asked for it instead of a different conversation — and sharing it with your learners works in a more natural way.

The bar at the bottom of your master now tells you exactly what's missing and takes you straight to it — intent, key points, and common errors are what make it ready, a reference piece is a bonus. And while you refine it, the AI now tells you plainly when tuning components further won't help and reviewing a real piece against it is the next step.

Browsing your master's pieces now gives you a "Craft a piece" action right at the bottom of the screen, with your master already selected.

Step connectors, master renames, and step reordering now save cleanly — no more false conflict errors when nothing underneath actually changed, and no more connectors showing as linked when a save silently failed. Live updates to a master, playbook, or project also stay connected through long sessions, and if someone else's edit does conflict with yours, you're told exactly who made it.

Key-point labels no longer overlap when several sit close together, the arrow-tip handle now appears reliably on hover without drifting, and none of this editing chrome shows up in exported images.

Text areas now expand to fit what you write in every browser, including Safari and Firefox, where they used to freeze at a fixed height and force internal scrolling.

Deleting a person, a project, a skill, a piece — anywhere across your workspace — now takes effect the moment you confirm it, with no wait for the page to catch up. If anything goes wrong, it quietly restores itself.

Creating a playbook inside an organization now updates the sidebar right away instead of leaving the new playbook missing until you refresh.

Key points, steps, common errors, review steps and their questions, exports, home sections, and training collections all now disappear the moment you delete them, and reappear automatically if the delete fails.

Members without a seat on every playbook no longer see the full org list on their home grid — only the playbooks you're actually seated on appear, matching what you can open.

Resend and cancel are working again for cohort invitations. Publisher organizations also no longer see another organization's pending invitations.

Your organization drawer now opens straight to what matters — change a role with a single click, resend or cancel a pending invitation inline, and copy an email address from one clear spot. The old per-row menu is gone.

Images generated through your connections now come from a stronger model, with the same reliability if one image in a batch fails.

Edit an earlier message and everything that followed is cleared away, so you can restream from that point on a corrected course — including restoring your piece to where it stood at that moment, when needed.

It now fades away the instant your answer starts, instead of lingering as a stuck timer on a finished conversation.

After you edit a piece by hand, your coach responds to what you actually wrote instead of what it last generated. On mobile, its replies now stream in live too, with a thinking indicator and follow-up questions appearing as they arrive — matching what you already get on the web.

Saving your work no longer throws a phantom "another editor saved first" error — genuine conflicts with a second editor still surface exactly as before. Selecting a table or block and pressing Cmd+C now copies it straight away, with no browser prompt asking you to copy manually.

When a background job fails, that failure is now recorded instead of vanishing without a trace.

After you hand-edit a piece yourself, your coach now recognizes the change as yours instead of describing it as something it wrote.

Editing intent, reference, introduction, or template past your first save no longer locks the field with a false conflict, and the "live updates unavailable" warning no longer interrupts you when opening a master — updates simply resume once the connection is back.

A transient network issue while creating a master used to surface as a confusing validation error. It now reports plainly as a temporary service issue, and a master is never left half-created when that happens.

The connector picker in step editing now shows only for organizations with connectors enabled, and loads your custom connectors only when you actually open it.

Opening an org playbook you haven't joined now takes you straight to the join dialog instead of the ordinary reading view.

Invitation outcomes now show the playbook's actual name instead of its internal ID, and re-inviting an existing learner to a higher role correctly raises their access rather than being silently ignored. The invite dialog now completes cleanly for every outcome.

Your organization's introduction now appears once — editable if you manage it, read-only if you don't — with the stray image sliver and empty strip that used to show around it gone.

Images embedded in markdown now load in batches instead of one request at a time, so pieces with several images appear faster.

Files and images get more time to load before giving up, so a momentary slowdown no longer shows up as a failure. Navigation also now takes you straight into your master or playbook, instead of routing through an editor-preview link.

When a review or refinement hands you back to the composer, your cursor is ready to type immediately, and any file that just attached gives a brief visual pulse so you know it landed.

The prefilled text after finishing a review now names your attached PDF directly and points you toward refining your Review Steps — the part that shows your actual review method.

The Improve card now stays out of your way when there's nothing to refine yet, and gets out of it once you're working: the concern field is focused the moment it opens and grows as your notes get longer, and your analysis scrolls itself into view the instant it's ready.

Analysis of your master now reads as clear, structured prose that names what's affected instead of pointing at an index, with proper formatting and paragraph spacing throughout. While aïno studies your piece, the text reads as the real content it is, and a pulse — not a spinner — signals it's actively at work.

If you're invited straight into a learning role, you now land directly in your learning space instead of being asked profiling questions first.

When a role change is blocked — by a publishing limit or because someone is the last owner — you now get a precise, correct reason instead of a generic error.

Inviting someone now walks you through two steps — what you intend, then a full preview of the consequences: playbook grants, member upgrades, and the role each person will hold. Each outcome is spelled out in plain language, with the role and playbook names bolded for quick scanning, and everyone who ends up with the same outcome is grouped into a single line rather than repeated one by one.

The preview now stays inactive with a hint until you've picked a playbook, and the role section is labeled THEIR ROLE to match what you're actually setting for the people you invite. Choosing Member or Admin for a guest no longer resets itself, the button reads "Invite" rather than "Send," and the sharing screen tells you plainly to reach out however you already do instead of implying email sending is coming.

Guest invites are now blocked whenever a person's actual access on any playbook — including shared ones — would exceed Guest, and previewing a batch requires at least one playbook selected first. These cases are caught right in the dialog instead of producing a confusing result later.

Removing someone from your organization now clears their past Learn, Craft, Train, and Design conversations for good — re-inviting the same person by email starts fresh instead of resurrecting old dialogue.

Your home now opens with a single, consistent view tailored to you — your progress, your playbooks, and, if you manage an organization, a summary of your people. It names the space you're in, and if you land somewhere empty while your real work sits elsewhere, a direct link takes you back. Click the prompt and your cursor lands right in the conversation, ready to type.

The cover image now sits at the top of the page, your space name no longer repeats itself beneath it, and your headline and next steps are centered rather than crowded together. Stray profiling prompts under an organization's introduction are gone, the welcome headline no longer shows twice while you're getting set up, and your personal space now displays correctly translated wherever it's listed.

Unpublishing a partner now clears them from your Partners list right away, instead of lingering until you reload the page.

Admins now show "Organization-wide access" instead of misleading enrollment labels like "Not started," so their standing is never mistaken for stalled training. The action menu is simpler too — the "Copy invitation link" option is gone, since invitations go out through your own channels.

Markdown files opened from the AI sidebar now render as proper formatted text — headings, bold, lists, tables — instead of a plain preview. And copying text from a learner view, an older master version, or any read-only piece keeps that same formatting instead of pasting as stripped plain text.

Pieces avoid the tics that give AI writing away — repeated rhythms, hedging phrases, and constructions like "not X, but Y" — so what you get back reads more like something a person wrote.

Generating a piece with your AI no longer dumps the whole document into the chat bubble — the conversation shows your exchange, and the piece lives where it belongs.

Pieces and reviews built mostly or entirely around an image no longer turn into raw markdown text when you reopen them — your images display correctly whether you're editing or just viewing.

If a file can't be retrieved, your coach now tells you promptly instead of leaving the screen frozen.

The button that puts you in your learners' seat is now labeled Learner view, with an eye icon that says what it does.

Starting a piece review no longer stalls partway through or dumps its findings as plain chat text — your coach now walks the review to completion, framed as a focused read of your work. Findings appear as structured review steps and questions you can act on, and adding a question to a step you just created works right away, without errors.

New and moved masters now keep their place within a section instead of jumping to the top — ordering holds whether you're creating, moving, or reorganizing them.

Accessing a playbook you're already part of now activates your reserved seat as it should, instead of leaving it stuck in reserved status.

Multi-step document generation no longer stalls into apologetic chat text — your edits are acknowledged properly so your piece keeps building turn after turn.

The invite dialog now labels organization membership and playbook access as separate, clearly named choices, so you know exactly what you're granting before you send it. Behind the scenes, invitations now hold to the role ceiling you've set, and role changes, bulk invites, and simultaneous acceptances no longer cost or grant seats by accident.

Guests and members no longer see empty, unusable sections cluttering the org home — only admins, who can actually fill them, see them at all.

Crafters see craft and pieces actions, editors and owners also see knowledge, and learners get a clean view with none of the authoring controls in the way.

Chat with your coach now holds together end to end: replies read as one clean message per exchange with tool activity folded in, proposed actions like review steps register correctly instead of stalling, and repeated plans no longer surface as duplicate cards. A rare pairing issue that could make a conversation unusable is also fixed — any conversation affected before now repairs itself the next time you open it.

When inviting people in bulk to a playbook shared from another organization, roles are now capped at whatever level the publisher allows — no more accidentally granting editor or owner access on shared content.

As an editor, you can now switch into a learner's view with a persistent banner marking the mode, and a link back to your editor tools whenever you're ready. From either view, the breadcrumb at the top tracks and highlights the section you're actually reading as you scroll through a master.

Learning a master now moves through a clear arc — destination, essentials, pitfalls, practice, craft, and a final step to share your thinking with your coach — so you always know what comes next.

Your cover image now sits in line with the rest of your content, at the same width, and can be removed like any other section. Sections you add from the popover now scroll reliably into view, centered and clear of the floating toolbar.

Add a section anywhere — your org, your personal space, a playbook — and it now scrolls into view with a brief highlight, so you never lose track of what you just created. The action bar has also been fixed to stay pinned to the bottom, even when a cover image hides the header.

Inviting people is now a single composer that takes you from who to invite, to their membership, to the playbooks they'll use — including clear labelling and attribution on playbooks shared in from elsewhere. Every invitation now shows its playbook correctly on your team's People tab, and cancelling works reliably no matter what kind of invitation it is.

Once your membership is removed, a stale invitation can no longer let you back in. And trying to reach a master you don't have access to now looks identical whether it exists or not — private content stays private, with no hints leaking through.

The membership explainer now correctly names your organization instead of leaving the reference blank.

Organization admins appear as themselves in the people list, even without a playbook seat — and your collaborator counts no longer double-count someone who is both an admin and a seat holder.

Your home page, personal space, and playbook editor now scroll as a single page — the cover moves with your content instead of staying fixed while a cramped inner scrollbar does the work — and a cohort playbook view that was clipping content past the bottom of the screen is fixed too. Create a master or playbook and it now scrolls into view with a brief highlight, so you always land on what you just made.

Opening aïno now unfolds like a title page — the mark settles in, the wordmark fades up, and a quiet grace note welcomes you in, before the app itself appears.