Lore Extraction, Plot Grid Trust, and a Mobile Keyboard Fix -- Plotiar Product Updates (July 2026)
There is a particular kind of dread that comes from a screen that says "Saved" when you are not sure it means it. I watched a friend restart her laptop mid-scene once, trusting the little checkmark that told her everything was fine. It was not a Plotiar story, but it is the exact fear I think about every time we touch anything that sits between a writer and the words already on the page. The last two weeks of work were mostly about that trust -- making sure the app tells you the truth about what happened to your work, every single time, even in the unglamorous cases nobody notices until it goes wrong.
Lore Extraction Now Handles a Full Manuscript
Pulling lore out of a long document used to hit a wall the moment the document got long enough to matter -- a full novel, not a chapter. The job could run out of room mid-answer and come back reporting "No new lore entries detected," which looked exactly like a genuinely empty result even though it was not one. Extraction now runs against limits sized for a real manuscript instead of a short excerpt, and it streams its answer so a long generation does not get cut off partway through.
Two related annoyances are gone too. A job that got stuck mid-run used to just sit there indefinitely; it now recovers on its own instead of spinning forever. And if your extraction actually succeeded but the result arrived a beat after the status flipped to "done," you used to see a false failure toast and get bounced back to the start of the dialog. That race is closed. A successful extraction now reads as a success.
Plot Grid Saves Tell You the Truth
A writer reported a plot grid that showed "Saved," then came back empty on the next visit. Nothing was actually lost -- the data was intact the whole time -- but a brief connection hiccup made the app claim success when the write had not gone through, and the next load rendered a blank grid in its place. That is about as bad as a false alarm gets for a planning tool. The app now waits for real confirmation before it says "Saved," shows an honest "Error saving" message and retries automatically when a save fails, and it will not silently hand you an empty grid and let you mistake it for your own work.
Tapping to Write on a Phone Actually Opens the Keyboard
A frustrating one for anyone drafting on a phone: tapping into the editor sometimes did nothing. No cursor, no keyboard, just a canvas that looked interactive but was not listening. It traced back to how the editor first wires itself up on a touch device, and it did not always finish that handshake before your first tap landed. Fixed -- tapping into a document on mobile now reliably brings up the keyboard, whether it is your first open of the day or your fiftieth.
Smaller Fixes Worth Knowing About
Document export had a stretch where it failed outright rather than producing a file; that is resolved, and exports are reliable again. If your plan includes bonus AI credits, a billing bug meant those bonus credits could get counted twice against your weekly limit, which cut some people off with real credits still sitting unused. Credits now draw down in the correct order -- your weekly allowance first, bonus credits only after that is spent -- so you get the full amount you are entitled to.
"Saved" should be a promise, not a hope. That is the bar for everything that touches your manuscript, and it is the one we keep coming back to.