Skip to main content
← BackChangelog · edition by edition

What changed, in plain ink.

Every edition that's landed on the canvas — brand work, reliability passes, security patches, the boring bits and the ones that felt like a corner turning. Most recent first.

Faster canvas navigation at low zoom.

Canvas performance when zoomed out is now significantly faster. Connection lines remain clearly visible at 30 percent zoom, with render speeds up to ten times faster.

Methods on the canvas now maintain their state consistently during renders, removing unexpected visual gaps.

The credits you paid for now stay.

Purchased credits no longer expire alongside your monthly allowance. A refund now returns to the pot it was taken from, so a run that fails gives back exactly what it cost.

Model settings tell the truth as well: the list names only what a run can actually reach, and a premium model shows its surcharge before you start rather than on the bill.

Send canvas cards straight to Linear.

Export design cards directly into Linear to turn canvas ideas into tracked issues without leaving your workspace.

Seamless exports and sharper cards.

Exporting canvas stories to project tools like Linear is now seamless. Issue types, priorities, and project mappings sync cleanly without duplicate exports.

Competitor and Glance cards feature updated layouts and clearer typography, making your research far easier to digest.

A quieter workspace and clearer forms.

Connection forms now explicitly highlight required fields so you can configure integrations with confidence. Your usage metrics also focus strictly on the projects you own rather than every canvas shared with you.

Release announcements wait until you open them, and signing in with a third-party identity provider now works across a broader range of OAuth services.

Fewer steps between thought and method.

Drag files straight onto the canvas to create matching methods instantly. You can now pull individual scenes out of script cards, navigate context menus entirely from the keyboard, and share live views that behave just like the main editor.

Method connections pass data seamlessly across your canvas, annual subscriptions are ready to purchase, and offline edits stay safe until you reconnect.

Methods that ask first.

Nothing runs until you have said what it should run on. A method that draws on more than one source lets you choose which sources feed it, and which individual items inside each one — leave a question out and it stays out of the brief. Before a run starts, the method checks the inputs it will read and names anything still missing. Each method carries its own model picker, with the price beside the choice.

A run is something you can watch: stop it mid-flight, read a brief of what is about to happen, and whatever you placed or edited by hand stays where you put it. When a method needs something clarified, it asks you the question and waits for your answer. The library grew alongside — Epic cuts work into themes and shows its reasoning, and the media methods round it out with generated images, audio and music composed through Lyria.

Every method, exported as itself.

A canvas now leaves Plooi as slides drawn in the shape of the method behind them. Over twenty methods export as themselves: a SWOT quadrant stays a quadrant, a Dave Gray empathy canvas keeps its six fields. In PowerPoint these are native shapes, not flattened pictures, so they can still be moved, recoloured and edited on the other side. The brand fonts travel inside the file, so print and PDF match the screen offline.

The same stretch also opened the front door. You can sign in with Google or a password, with the requirements checked as you type, and the waitlist asks for consent up front, then confirms by email. Approval arrives as an invitation; a decline arrives as a kind note that names the date, thirty days on, when you are welcome to apply again.

Twenty methods, drawn as methods.

The method library roughly doubles. Twenty new methods land — Stakeholder Analysis, Customer Journey, Jobs To Be Done, Business Model Canvas, Crazy 8s, Storyboard, Elevator Pitch and thirteen more — and nine that were already there were drawn again. The point of the pass is the shape: a method now looks like the method. Stakeholder Analysis is a power and interest grid.

Underneath, every card reads at three sizes, keeps Run and Re-run in the same footer whichever method it is, and shows where its content came from. Three of them do their work on the canvas itself: Wireframes draws an image per screen at sketch or mockup fidelity, the Minimum Viable Product card lets you drag features between scope columns, and Storyboard gives every beat its own output.

What changed, and what it will cost.

The canvas now keeps track of itself. Edit a method and everything downstream of it is marked out of date, on the cards and on the connections alike. A chip above the zoom bar counts them and estimates the credits a re-run would take. Click it and the whole set is selected, with the viewport already on the first one. Nothing re-runs by itself — the decision stays yours.

Cost arrives before the spend elsewhere too: a cost card totals the credits for a full downstream path before you start. Two things shorten the trip across a big canvas — frames fold into a single card, and a keyboard palette travels to any method by name. Five recipes, from Discovery Sprint to Validation Loop, go down in one step that a single undo takes back.

Three methods, drawn on paper.

Three methods join the library. Empathy Map arrives in the Dave Gray layout — Think, Feel, Say and Do — and draws its own insights out of what it gathers. Decision takes several strands and lands on one call, every option carrying a confidence reading. Sticky notes go anywhere. And SWOT has the shape of the method itself: an empty two-by-two that fills, with every run kept, so an earlier version can be read back.

Every generated item carries one chip: who wrote it, how sure it is, why. Colour reads the confidence at a glance, and the reasoning opens on hover. Reach for a bullet and it turns into a lock — anything you touch by hand stays as you left it. App, marketing site and email settle at one colour temperature, cooler neutrals under the navy accent.

Plooi writes back.

Plooi now writes to you at the moments that matter. Add someone to a project and the invitation goes out; change their role and they hear about it; share a canvas and the link travels by email. Your account keeps its own side of the correspondence too — a welcome after the first sign-in, a confirmation when a password changes, a heads-up when credits run low or a payment needs attention, while there is still time to act.

You can see what a run costs before you start it: generate buttons carry their credit price, and the expensive ones ask once before they go ahead. Credits bought as a pack survive a renewal, a refund returns them the way they were taken, and a job that never completes gives them back on its own. Accounts can be protected with a code from an authenticator app.

A new sheet, folded open.

Plooi gets the brand it always wanted. Riso-blue ink, warm paper, Bricolage Grotesque headlines — and the chatbot-era visual debt left at the door. The marketing site is now twelve pages with the same vocabulary on every fold.

Highlights: a redesigned home with a draggable hero canvas, a "Why Plooi" page that puts the chatbox next to the canvas, a philosophy long-read with a working zoom slider, and a methods library you can actually search.

Email, finally delivered.

The waitlist invitation email is now a server-rendered template with a real Resend pipeline behind it, instead of relying on a browser to assemble the HTML. Cron retries will land in the same inbox.

As part of the same pass, the Tally webhook stopped leaking membership: duplicate signups now get the same neutral acknowledgement as fresh ones, so the webhook URL can't be used to enumerate emails.

CTAs that match the moment honestly.

The landing page used to say "Start with a sheet" while the product behind it was waitlist-only. The CTAs now route to the alpha waitlist by default, with a single feature flag flip (VITE_WAITLIST_MODE=off) to switch the entire site back to self-serve at public launch. Plus the German strings in the app shell got translated, the testimonials lost their fake company names, and the waitlist confirmation tells you what to expect.

Bring your own keys.

The Plooi canvas now runs AI methods against your own provider keys — Google Gemini, OpenAI, or Anthropic. Keys are stored server-side, encrypted at rest with AES-256-GCM, and never sent to the browser. Switching providers is a single dropdown; every method remembers what ran it.

The community plan is BYOK-only (bring your own key) by design. Paid plans keep access to the system Gemini key for users who don't want to bring their own.

A bug-audit marathon.

Fifteen parallel audit rounds, ~120 findings, ~95 fixed and deployed. RLS was tightened on public_project_links, bug_reports, and user_profiles. The Stripe webhook got atomic idempotency, the email sender stopped being an open relay, the password-hash endpoint is now cron-secret gated, and the AI proxy has a server-side throttle so a single tester can't drain the whole pool.

Migrations 20260526000000 through 20260527090000 shipped. Forty new tests landed alongside.

Subscribe? Not yet — the changelog is read-only for now. If you'd like to be told when something material ships, join the alpha and you'll get the same notes by email when they're worth your time. Older work lives in docs/ in the repo.