What's new

Follow what changed in Content Ninja with each update.

v1.6.0 · August 19, 2026

Added

**30% off large product-text runs, if you can wait a day or two.** When you rewrite 200 or more products at once, a green bar appears above the Generate button in the New run dialog: switch it on and the same text is written, just not in minutes but typically within a day or two — and it costs 30% fewer credits (200 products becomes 140 credits instead of 200). Your credits are reserved when you start, and the charge happens in a single item, only for the products that were completed. We email you as soon as it is done, and the **Shopgrade basics** assistant offers it when you discuss a large catalogue. [Kedvezményes feldolgozás: 30% kedvezmény, ha ráér 1-2 napig](/help/shopgrade-kedvezmenyes-feldolgozas)
**The discount now applies to products you add later — and you have 5 minutes to change your mind.** You do not have to select your whole catalogue at once: the **Product** button lets you add more to a run later, and every batch earns the discount on its own from 200 products (items that went in at full price do not become cheaper retroactively). After you start — and after every new batch — the header shows a **Cancel** button for **5 minutes**: it costs nothing and your reserved credits are released immediately. Once processing has started, **Stop** takes its place — from then on the run cannot be deleted, but you can review and approve the finished texts at any time, and we write them to your shop. If you do not have enough credits, the addition does not start. [Kedvezményes feldolgozás: 30% kedvezmény, ha ráér 1-2 napig](/help/shopgrade-kedvezmenyes-feldolgozas)
You can now see **where one of your product pages stands** — via the **Product page diagnosis** button at the top of the Shopgrade page. Paste a link to one of your products and within seconds you get a 0–100 score for how search-friendly it is, how quotable it is for an AI answer engine, and how well it helps the customer decide — plus a list of what is missing. It is free, it writes nothing to your shop, and your earlier diagnoses are listed there too. The **Shopgrade basics** assistant can run it in chat as well: it writes your prompt from the measured gaps instead of generalities, and after a trial run it tells you **with a number** how much the text improved. [Terméklap-diagnózis: mérd meg, hol tart a boltod](/help/shopgrade-termeklap-diagnozis)
The product page diagnosis now measures **four aspects**, not three. The new one, **Consistent and on-brand**, is built from what you answer: how consistent your product pages are, and how much they speak in your own brand voice — this cannot be measured from a single page, yet it is exactly what one run fixes. The questionnaire therefore gained a new step (where your current text comes from, how consistent it is, how on-brand it is, and whether you have a supplier price list with extra data), and we **ask signed-in users too**: in the Product page diagnosis dialog and with the **Shopgrade basics** assistant. If your text comes verbatim from the manufacturer, we don't even ask the brand-voice question. Anything you skip **does not count against you** — that aspect is simply left out. We also look at **formatting** now (subheadings, tables, highlighted blocks), but strictly inside the description itself: your shop template's spec table earns no points. The maximum is still 100.

Fixed

**Longer prompts can now be saved**: the prompt text limit grew from 5,000 to **12,000 characters**, and a counter under the field shows where you stand (e.g. 4 312 / 12 000), turning red if you go over. Until now a thorough Shopgrade prompt hit the old limit and all we said was “Missing or invalid data”, even though every field was filled in. **Ninja AI** is now told which field is at fault too — so it shortens the text by itself instead of retrying the same thing; this applies to every AI action's error message, not just prompts. New help section: How much fits. [Prompt-írás alapok](/help/prompt-iras-alapok)
Asking **Ninja AI to create a newsletter** (or a blog article or an ad) now works exactly like creating one with the button yourself. Until now it was only half-created, so the newsletter stayed empty and Generate stubbornly said "A generation is already running for this content" — even when nothing had started, and reloading the page didn't help. On top of that, the newsletter now gets your **design template** and the default recipe, and the link in the AI's reply takes you to the **newsletter editor** (instead of Posts). Older items created this way heal themselves as soon as you save or generate on them.
Reconnecting Facebook no longer silently kills your other connections. Facebook only keeps the pages you ticked during your **most recent** sign-in — the ones you skip stop working, even if they belong to another brand or another workspace. Until now this only surfaced **the next morning**, and scheduled posts didn't go out in the meantime. Now we tell you how many pages are already connected when you start (tick those too), and on your return we **list right away** which connections stopped — by name in your own workspace, as a count for the others — with a button to restart the sign-in. Rows that are already connected get a new **Refresh connection** button that saves the fresh access to the account without breaking the assignment. [Miért állt le a Facebook-kapcsolatom egy újrabekötés után?](/help/facebook-kapcsolat-leallt)
When your sign-in expires, we now **say so in plain language**. Until now you got an unreadable English sentence (or nothing at all) and had to guess what happened — now we tell you the session expired and you need to sign in again. If it wasn't the session but simply a dropped connection (the machine went to sleep, the tab was backgrounded), we say that too and ask you to reload the page.
The **Visual step warns you up front** when there is no image to put the logo on. Until now, if a **video** was selected on the left (or an image that was still being generated), generation started and only then turned out that the logo step had no input image. Now it asks before generating, and it states **exactly what the situation is** — that a video is selected, or that the image is still being generated — instead of a generic sentence.
If your newsletter provider **rejects your sender address**, the warning now **stays next to the Sender email field** until it is resolved. Until now the error appeared after sending, in the Status block — easy to miss, and easy to run into again. Providers only validate the sender at send time, so we cannot prevent it; what we can prevent is you hitting it a second time.
A **momentary AI outage no longer costs you the week's content**. Until now, if the AI provider stopped responding, generation died silently: the post stayed empty and the automation closed with a meaningless error ("Generation did not finish for one of the contents"). Now the system **retries by itself** — immediately first, then up to twice more in the following rounds — and in most cases it resolves without you noticing. If it still fails, the **run log tells you which content and what** got stuck, and the content stays a draft so you can restart it with one click. The same protection covers image and video generation. Updated help section: When something fails. [Automatizmusok beállítása](/help/automatizmusok)
Explanations now open on phones and tablets. The "?" icons and other explanatory bubbles only appeared on hover with a mouse — tapping them on a touchscreen did nothing. Now a tap opens them, another tap (or a tap anywhere else) closes them, and the "?" icon has a larger touch area so it is easy to hit with a finger.

v1.5.0 · August 14, 2026

Added

New automation trigger, **Low keyword stock**: when fewer than the given number of approved keywords remain in a research, a new keyword research starts on its own.
Blog posts can have templates too: set the recipe once (tone, knowledge source, model, images, cover, target menu), and every article — manual or automated — is written with it.

Changed

The blog step in automations is now as simple as the post step: it only asks for a template — brand, blog account and topic all come from it.
You now decide which menu item of your store your blog articles are filed under. Your blog account has a new field (**Hova kerüljenek a cikkek?** — where should the articles go?) listing your store's actual menu items: by default we offer the one where your existing articles live, marked **(javasolt)** (suggested), but you can pick any other. The article's **Publikálás** (Publish) step has a **Hol jelenjen meg?** (Where should it appear?) field too: you can see where the article will land, and you can override it for that one article — a seasonal menu, say — while the account's setting stays as it was. Until now the system worked your blog menu out on its own, which was right for most stores — but if you had no published articles yet, it had nothing to go on, and the article was left out of the store's blog listing. That can't happen any more. The detection got more accurate too: we suggest the menu item where **most** of your articles are, so one article forgotten in an old, abandoned menu can't skew the choice. (UNAS stores; Shoprenter has a single article list, where articles land in the right place by themselves.)
The automation editor's **Action** menu now only offers steps that actually make sense. Until now you could build a rule that quietly did something other than what you expected: a date set alongside immediate publishing never took effect, and topping up the keyword set next to a post only burned your credits. Those steps no longer appear where they don't fit; the ones blocked only by order (such as **Schedule it** without a date) stay visible, greyed out, with the missing piece spelled out underneath. You can still create several pieces of content in one run — you just set their type in one place. Nothing is removed from your existing automations: if one of them still contains a step that no longer takes effect, that row gets a yellow warning and you decide what to do with it. New: you can also delete the **Create…** step, so you can build an automation that only keeps your blog topic set stocked.

Fixed

Analytics no longer goes quiet when data is missing. It could happen that the Facebook view sat empty — no numbers, no posts, no warning — even though nothing was wrong with your account: it was enough for Facebook to retire a single metric, and that wiped out the statistics of every one of your Facebook posts, likes and comments included. From now on such a dropout only takes that one number with it and the rest of the data still arrives; and if an account's daily figures stall, we still fetch your posts' statistics. If something is missing anyway, we say so: the "stale account" bar now names the reason too (an expired connection, for example), and a separate, quieter notice tells you when the account's figures are fresh and only a few posts' statistics haven't arrived — the next daily refresh picks those up. A new help section explains what these bars mean. [Analitika: hogyan teljesítenek a posztok](/help/analitika)
**Generate from template** now actually starts. It could happen that the post was created, the editor "thought" for minutes, then said the generation never started — and it really hadn't: the work was meant to be kicked off by your browser in a separate request, which could get lost while the editor was opening, and then nothing happened on the server at all. From now on creating the post and starting the generation happen in a single step, and the work carries on server-side even if you close the tab. We also warn you in advance when a template's visual recipe won't produce an image: if the recipe would start from an existing picture (Logo, Animation or Resize, with no image generation), there is nothing to work on in a freshly created post. You see this in the template editor, in the template picker and in the campaign settings before you start it — you can still run the generation, the decision is yours. The same warning appears on a post's **Generálás** (Generate) button when the gallery is empty. [A sablonból csak szöveg készült, kép nem — miért?](/help/sablonbol-nem-lett-kep)
Your automation no longer skips a post in silence. It could happen that the automation wrote the text and created the image, yet the post never went out: it stayed a draft in the calendar, and the run log said "Error — No content was created" even though the content existed. The cause: while the text was being written, another background run started in parallel, saw the work as unfinished and closed it. From now on nothing else can touch a run in progress, and we only say that no content was created when that is actually true.

v1.4.0 · August 14, 2026

Changed

Your brand's look now reaches everything on its own. **Visual analysis** is switched on by default both when you add a brand and when you re-analyse one, and it now reads your store's stylesheets as well — so it finds your actual colours (with exact colour codes), your typeface and your shapes, not just the handful of colours visible in the homepage markup. The **Visual style** field became short and factual as a result: three lines that every module reads the same way. Image generation, newsletter templates and Shopgrade's styled product descriptions all use it (and your AI influencers get the same three lines) — and for the latter you **no longer need to paste colour codes into your prompt**, it uses your store's colours by itself (a colour named in the prompt still wins). If a re-analysis would overwrite the style you fine-tuned by hand, we warn you first.
Your monthly Ninja AI budget now stretches much further: a ten-question conversation costs a fraction of what it used to, while the answers stay the same — in fact it remembers the recent part of the conversation more precisely. And if the budget does run out, the chat no longer goes silent: you can still ask questions and it can still look at your data, it just won't perform actions (creating posts, editing, publishing) — instead it shows you where to do it yourself in the app. We warn you at 80% of the budget, and when a conversation gets long, the chat offers to start a new one for your next topic. [A Ninja AI havi kerete](/help/ninja-ai-havi-keret)
The free product page diagnosis now starts with your email address: you tell us where to send the result, and the evaluation runs right after. If you arrived from our website's form, the field is already filled in — just hit Next. In return your result no longer gets lost: if you gave us your address but didn't run the diagnosis, you get a reminder the next day that picks up where you left off; if it ran but you didn't ask for the coupon code, you get a message with your own score and why your weakest point is a problem for you, and two days later one about what happens if you leave it as it is. Three messages at most, and as soon as you run the diagnosis or ask for the coupon code, the remaining one is dropped automatically. Every message has an unsubscribe link at the bottom. A shared result page still opens without any email address.

Fixed

Content is no longer created while we silently skip something. If we can't work with a setting you switched on — no knowledge source selected, no link on the post, or no image tool chosen — we now ask before **Generate**: **Beállítom** (I'll set it up) or **Mégis generálom** (Generate anyway). No credit is spent until you decide. If an automation or Ninja AI created the content, the warning stays at the top of the post (with a ⚠ marker in the list), and the run log shows a **Részleges** (Partial) row with the reason. Image generation from a product link got better too: we now also use the link written **in the post's text** (until now only links saved during generation), so posts brought over from the old system or written by hand get your actual product on the image instead of an invented one — and if the link yields no image, we tell you that as well. If a generation gets stuck, after three minutes we say it didn't start so you can try again. [Hiányos generálás: mit jelent a figyelmeztetés](/help/hianyos-generalas-figyelmeztetes)
A webshop knowledge source brought over from the old system now really comes to life the moment you connect your store. Until now the migration carried over every setting of the source but — for security reasons — not your store's key, and nothing tied the two together: the source silently returned nothing, and your automations produced generic posts instead of ones about your products. From now on, saving the connection links your waiting sources automatically (the confirmation even tells you how many), and it works the other way round too, if the store was connected first. If you have several stores we don't guess for you: the source visibly stays "Waiting to be connected" and you pick its connection. On top of that, a misconfigured source is no longer silent: it gets a badge in the Knowledge sources list, and a red warning next to its name when generating, telling you what's wrong and where to fix it.
The Publish step of a post now only lists the accounts we can actually publish a post to: your Facebook page and your Instagram. Until now your newsletter and blog accounts showed up there too, marked "Not connected" — and if you ticked one, publishing got stuck on it. You now see the same list here as on the Text step. Newsletters are still written under Newsletters, blog articles under Blog articles.

v1.3.1 · August 11, 2026

Fixed

The free product page audit now actually finishes. On launch day every audit stalled at the evaluation step and spun forever: the one-minute budget for the background work was too short, and the system cut it off just before the result arrived. The budget has been raised, and the evaluation now watches its own clock — if it would run out of time, it closes with a partial result instead of stalling. If an evaluation does break off, we tell you within a few minutes instead of spinning on, and the message now points to the right next step ("start a new one") rather than claiming we cannot find your audit.

v1.3.0 · August 11, 2026

Added

Shopgrade can now write styled product descriptions: a coloured header box, feature cards and a spec table with a header row, instead of a wall of text. There is no new switch for it — your prompt decides: ask for it and you get the design, leave it out and everything stays the plain copy you have today. There is a ready-made built-in prompt ("Dizájnos termékleírás") with a single block at the top where you swap in your shop's colours; your existing prompts are untouched. The styling only applies inside the description: it cannot spill over to your shop's header, menu or basket. On the approval screen the Formatted view is view-only for styled descriptions — you edit them in the Source view. The Shopgrade guide asks once whether you want a styled description, and asks for your colours. [Dizájnos termékleírás: színes dobozok és kártyák](/help/shopgrade-dizajnos-leiras)
Free product page diagnosis: on the `/shopgrade-audit` page you can answer three short questions without signing in, drop in a product link from your shop, and within seconds get a 0–100 score for your current copy. We break it down across three criteria — search-friendly, AI-friendly and sales-focused — and for each one we spell out the goal and the yardstick: a green tick for what's fine, a red cross for what's missing, and quotes from your own text word for word. We also estimate how far the page could be lifted — where there are no facts to work from, we don't promise more. If you tell us your descriptions come from the manufacturer or your wholesaler, we factor that in too: the same text runs in dozens of shops, and search engines see it that way. We show what share of your description would be true of any similar product. The quote uses the fastest possible run and the most favourable scenario: what one product's copy costs, what the whole thing costs (subscription minus the coupon discount, plus any extra credits, itemised) and when it will be done — plus everything else the same subscription covers and what is coming soon. At the end there's a 50% coupon code that lives for 3 days from the moment you ask for it and covers 750 product descriptions. We never write anything to your shop, we only read the page. [Ingyenes terméklap-diagnózis: mennyit ér most a szöveged?](/help/termeklap-diagnozis) [Mit jelent a cserélhetőség a terméklap-diagnózisban?](/help/mi-a-cserelhetoseg)

Changed

A scheduled post can no longer stall silently: if a publish run is cut short, the system picks it up again within minutes and continues where it left off — anything already published is never sent twice. Until now such a post could sit in "In progress" forever with nothing to flag it. Scheduled posts now go out every 5 minutes (up from 15), and Instagram images usually publish on the first attempt instead of minutes later.
Shopgrade now also writes **where a fact comes from**: when a detail carries a standard, a certification, a measurement or the manufacturer's data sheet, the text names it (for example "eye protection to EN 166 F") — it still never invents anything. The list of facts stays neutral, too: no "premium" and no "high quality", because a praise adjective takes away exactly the credibility that would make an AI answer engine quote it. It only writes a question-style subheading where the product's own data can answer it — it never asks an empty question. The product page diagnosis follows suit: with no facts to work from it no longer promises the points a rewrite cannot earn, and on a weak score it says that a shop can appear in AI answers even when it isn't near the top of Google's results.
The Shopgrade guide (Ninja AI) now covers the things that can undo your whole job: before writing the prompt it asks whether your products are refreshed automatically from somewhere (a wholesaler connection or a sync tool), and tells you what to take out of the sync so it won't overwrite your rewritten texts. It asks how it should address your customers and what to call your products — this goes into the prompt as a fixed section, so every run sounds the same. It explains the finished prompt in plain language before saving, and gives you a pace for working through your shop. It now reads the texts from your test run itself and suggests the fixes — you no longer have to paste them into the chat. [Automata termék-szinkron: hogy ne írja felül a munkádat](/help/shopgrade-szinkron) [Tegezés vagy magázás? Egységes szóhasználat a termékleírásokban](/help/shopgrade-nevezektan) [Milyen ütemben írasd át a boltodat?](/help/shopgrade-utemezes)

Fixed

Blog articles published to your UNAS shop now show up on your shop's blog page automatically — previously the article was created but you had to assign it to the blog list by hand in the UNAS admin. We never touch the placement of your existing articles, and republishing doesn't create a duplicate. Your shop's API key needs the page-management (getPage/setPage) functions enabled for this.

v1.2.0 · August 8, 2026

Added

For customers migrating from the old system: brands that were not linked directly to your subscription in the old system now come across too — until now such a brand could be silently left behind. The “What should we bring over?” step lists the brands by name, so you can see before you start that nothing is missing. If you already migrated and a brand was left out, we email you and bring it over with one click.
When a product fails, Shopgrade now tells you what happened: the red row opens up and explains the reason in plain language (including your store's own message where there is one). Temporary failures are retried automatically, and where retrying wouldn't help, we show what to do instead of a button. [Hibás termék a Shopgrade-ben, vagy nem tetszik a legenerált szöveg](/help/shopgrade-hibas-termek-es-finomitas)
If your product descriptions are empty, you can now upload your wholesaler's or manufacturer's product list (CSV or Excel) from the **Supplier data** button in the Shopgrade header — the AI then also works from the facts in that file (material, size, power). We tell you how many of your products found a match, and if the identifiers don't line up we offer a fix — we never rewrite a SKU on our own. When starting a run you can tick several files; on a clash your store's own data always wins. [Nagykeres vagy gyártói terméklista feltöltése a Shopgrade-hez](/help/shopgrade-beszallitoi-adatok)
Four new variables for Shopgrade prompts: Manufacturer, Category, Parameters and Unit of measure — you can insert them into your own prompts too. [Milyen hosszú legyen a termékleírás, és hogyan tagolódjon?](/help/termekleiras-hossz-es-vaz)

Changed

Shopgrade's rewrite buttons now do what their names promise: **Retry** tries again with the same settings (free when only the store write failed, since no new text is generated), **Refine** adjusts the current text — only the fields you tick, following your note — and **Regenerate** writes a new one from scratch. Until now a single button always rewrote everything, discarding your per-field choice and your note.
Shopgrade now also sees your products' facts (manufacturer, category, parameters, unit of measure) and writes more concrete descriptions from them — the real size, material and package contents go into the text instead of generic praise. The long description is now structured around the questions a buyer actually asks, and any section with no data is left out.

v1.1.2 · August 6, 2026

Changed

The editors are now comfortable to use on a small laptop too: the post text field no longer gets squeezed down to a couple of lines, and the Generate button stays in place while you scroll. On a short screen the editor window is taller and the spacing tighter — the text size is unchanged. On a large monitor everything looks exactly as before.
If your migration is still being prepared, you now get an email automatically as soon as you can start it — no need to keep checking the page.

v1.1.1 · August 6, 2026

Fixed

A new template (post, newsletter or campaign) now starts with the default text recipe already selected — so content created from the template knows your brand, just like a new post does.
Cover images for the built-in prompts are visible again in the editor's "Saved prompt" picker — until now it showed an empty placeholder there, even though the prompt list displayed them.
All your Facebook pages show up when connecting: we now guide you to the manual page selection at login (tick them all there), and if a page might still be missing we warn you with a "Reconnect" button.
A newsletter only goes out with the sender filled in — an empty sender used to fail at the provider. If something does go wrong, the alert email tells you which newsletter it was, what the problem is and what to do.

v1.1.0 · August 6, 2026

Added

Migrating from the old system: your automations now come across with their full setup (image or video mode, AI model, aspect ratio, logo and saved prompts), and we ask separately what should happen to your already scheduled content.
If your migration gets stuck, you can restart it yourself with one button — we continue from where it stopped, and nothing that already came over is copied twice.
After migrating, a welcome screen greets you: you see what came across, and a guided tour walks you through the new interface — you create a post together along the way, and at the end Ninja AI helps you connect your accounts. [Mi hogy hívják most? Kreatív tár, tömeges generálás, Cég](/help/mi-hogy-hivjak-most)

Fixed

Your sent newsletter keeps its formatting: headings stay in your brand colour and buttons no longer turn into underlined links — we bake the styles onto the elements themselves before sending.
Newsletter and blog post generation works again with all three AI models — previously the Generate button stopped with an error.

v1.0.0 · August 3, 2026

Added

What's new page: from now on you can follow here what changed in the app with each update — in Hungarian and English.