AutoKeyWorder Guide: Every Feature on All 5 Platforms

50 images a day, 12 toggles in a popup, 5 platforms that each behave a little differently. The AutoKeyWorder extension looks like a cockpit until you know what every control does, and this is the guide we wish we’d written for every new user. Full disclosure: we built AutoKeyWorder.

You install the extension, land on your first upload page, open the popup. A credits counter. A Context field. Two processing buttons. A toggle for Enhanced quality. Another one marked Overwrite existing titles. A dropdown that only shows up on Zedge. A checkbox for AI disclosure that sometimes appears and sometimes doesn’t.

If you skip past all that and just hit Process All, you’ll get metadata. It’ll probably be fine. But you’ll leave real accuracy, real credits, and real control on the table. Some of these toggles only appear on some platforms. Some cost you double credits. One of them quietly gates the entire difference between “AI wrote you a guess” and “AI wrote you the tags a buyer actually types.”

Here’s what every control in that popup actually does, across all 5 supported platforms, with the platform-specific behavior spelled out where it gets weird.

The 5 Platforms (Plus Two Bonuses)

AutoKeyWorder officially supports 5 platforms on our marketing pages and Chrome Web Store listing:

Platform	What you upload	Upload-form fields filled
Adobe Stock	Photos, vectors, videos	Title, keywords, category
Shutterstock	Photos, videos	Title, description, keywords, categories
Zedge	Wallpapers, videos	Title, description, tags, category, pricing
Displate	Metal posters	Title, description, tags, collection
TeePublic	T-shirts, merch	Title, description, tags

Two bonus platforms are also wired into the extension if you stumble onto them: Dreamstime and Etsy. Etsy even gets its own controls (digital product toggle, description tone). They’re live but we don’t lead with them in marketing yet. If you sell there, they work.

What AutoKeyWorder Actually Does

TL;DR: reads your uploaded images, asks a vision AI what they are, fills the platform’s metadata form. You review and submit. One credit per image (two for enhanced).

AutoKeyWorder is a Chrome extension that reads the images you’ve uploaded to a creator platform’s upload page, sends them to a vision-language AI model, and fills the platform’s metadata form with the AI’s generated title, description, keywords, tags, and category assignments. It never submits the listing for you. You always review every filled field on screen and click the platform’s native publish button yourself. Each of the 5 officially supported platforms has its own dedicated AI prompt file tuned to that platform’s search algorithm, category taxonomy, character limits, and buyer language patterns. This is why Adobe Stock titles come back reading like photography captions and Displate titles come back reading like decor product listings. Same source image, different buyers, different metadata language. The extension runs entirely on AutoKeyWorder servers, you don’t bring your own API key, and you pay in prepaid credits that never expire.

When you open the popup, the Detected badge at the top shows which platform the extension resolved. If it shows nothing or the wrong platform, reload the upload page.

The Controls That Appear On Every Platform

Start with these. They’re the universal toggles, same behavior regardless of which platform you’re on. Then we’ll get into the platform-specific weirdness.

Credits Bar

The digit counter at the top shows your remaining credits. Credits never expire. Clicking Buy opens either the credit-pack panel (300 for $4.99, 1,000 for $10.99, 5,000 for $49.99) or the subscription panel (Starter 750/mo, Pro 2,500/mo).

If processing fails on a specific image, the credit is refunded automatically. The backend refunds on any API-side exception, and the extension triggers a separate refund if the AI succeeds but the form-fill crashes afterward. Either way, you don’t pay for empty fields.

Enhanced Quality Toggle

This is the one most new users get wrong. When visible, the Enhanced quality toggle switches AutoKeyWorder between a standard AI model that costs 1 credit per image and an advanced AI model that costs 2 credits per image. The standard model is the default state of the toggle and delivers fast, high-quality metadata that handles the majority of straightforward stock, wallpaper, and merch subjects without much editing. The advanced model adds richer vocabulary, more nuanced keyword selection, and sharper editorial language, at exactly double the credit cost per image processed. On a 50-image batch that is 100 credits instead of 50. The decision is not about raw quality, it is about marginal benefit: whether the specific batch you are running now contains subjects where the advanced model’s extra vocabulary actually justifies the doubled credit spend.

Twice the cost for enhanced mode. That’s not a rounding error, it’s a deliberate trade-off you need to make every batch.

Turn it on when the subject is subtle, composition-dependent, or editorial. Macro detail. Abstract concepts. Niche editorial work where one missed keyword buries the listing. Fine-art illustrations with specific art-historical references. Editorial photography where the scene context matters more than the object count.

Leave it off for volume runs of straightforward stock subjects. Lifestyle shots, product photography, food, business scenes. The baseline model handles those well and doubling the credit spend doesn’t move the needle enough to justify it.

Real talk: on a 50-image batch, enhanced costs you 100 credits instead of 50. If 30 of those are standard lifestyle and 20 are tricky, run them in two passes. Select the 20 tricky ones, flip enhanced on, Process Selected (where supported). Then process the rest with enhanced off.

Context Field

This is the most important field in the popup and the one almost nobody uses.

The Context field is a free-text input up to 200 characters that sits below the main toggles inside the extension popup and shows the placeholder text “e.g. location, concept, theme, style, mood…” until you type into it. Whatever you type gets prepended to the AI prompt as user-provided concept information before each image in the batch is analyzed. It applies to every image in the current run and persists on screen until you clear the field or close the popup. The Context field exists because the AI sees pixels, not intent. A vision model can identify the objects in a photo with high accuracy, but it cannot tell on its own whether those objects belong to a tourism shoot, a documentary assignment, a lifestyle campaign, or a product catalog. Two keystrokes of context disambiguate an entire batch and change the downstream keywords, title voice, and description language the AI generates for every image in the run.

The AI sees your image. It doesn’t see:

Where you shot it
The story or concept behind the batch
Your intended buyer
Editorial framing you want reinforced
Event names, sports, seasons, or cultural references the pixels alone can’t convey
Technical qualifiers you want propagated into keywords

A photo of a person holding a sign could be a protest, a celebration, a marketing mockup, or a political event. A product shot could be for e-commerce, editorial, or an artisan Etsy listing. Context collapses that ambiguity before the AI guesses.

Good Context entries:

Shot in Lisbon, Portugal, summer 2026, tourism concept
Thanksgiving dinner, multigenerational family, American home
Minimalist flat-lay, Scandinavian aesthetic, morning routine
Protest against housing prices, Berlin, documentary style
Autumn hiking gear product photography, outdoor brand aesthetic

Bad Context entries:

Image of a woman: the AI already sees her
Make good keywords: gives zero new information
high quality: keyword stuffing, not context

If your batch mixes themes, process in sub-batches with different Context strings. The Context you type right before clicking Process is the one the AI uses. Set it per run, not once per session.

If you ignore the Context field, the AI still works. It just does its job half-blind.

Mark as AI-Generated

Checkbox. When enabled, the extension ticks the platform’s AI-generated-content flag during form submission. This matters because Adobe Stock requires AI disclosure on generative assets, and listings without it get rejected or removed.

Here’s the catch most articles miss: this toggle is Adobe Stock only. On Shutterstock, the checkbox is hidden because Shutterstock doesn’t accept AI-generated content from contributors. On Zedge, Displate, and TeePublic, the concept doesn’t map cleanly to any form field so the toggle doesn’t fire. If you’re uploading AI art to Shutterstock, don’t. They’ll reject it.

Turn it on for any Adobe Stock upload that passed through a generative model (Midjourney, DALL·E, Stable Diffusion, Firefly, Nano Banana, Seedream, Ideogram, Recraft). Turn it off for photography, hand-drawn illustration, or pure composite work.

Overwrite Existing Titles

Checkbox. Label says “titles,” but in practice it’s a per-field overwrite flag that applies to title, description, and tags independently depending on the platform. On Etsy, for example, it affects all three. On Adobe Stock, the title always writes (no opt-out). On the others, it checks the flag before replacing content you already typed.

Use cases:

You pre-populated titles from filenames and want to replace them
You’re re-running a batch with a new Context and want fresh metadata throughout
You renamed your concept mid-batch and want consistency

Leave it off when you’ve manually written metadata you want preserved, or when you’re only running AutoKeyWorder for the fields you haven’t touched.

Allow Brand/Trademark Names

Checkbox, off by default. This is a prompt-level filter. When off, the AI actively strips brand names, trademarks, and copyrighted properties from the generated metadata at generation time. Nike becomes “athletic shoes.” Starbucks becomes “coffee shop.” A visible iPhone becomes “smartphone.” The notice under the toggle says: “Brand names are automatically avoided to prevent copyright issues.”

This is the safe default. Adobe Stock and Shutterstock both reject assets with trademark-referencing metadata unless you have a property release, and the rejection explanation usually doesn’t name the trademark, so you have to guess which keyword killed the listing.

Turn it on only when:

You’re uploading editorial content to a platform that permits branded references
You have a signed property release
You’re on an editorial-first surface

For Displate and TeePublic, where fan art and licensed references are culturally common but still legally exposed, keep it off unless you hold the rights. The AI will find perfectly rankable synonyms.

Process All vs Process Selected

The two action buttons at the bottom of the popup. Here’s where platforms split.

Process All runs every image currently visible on the upload page through the AI pipeline in order, regardless of which platform you are on. It works on every supported platform and is the button you will reach for on most runs when the whole batch shares a single concept. Process Selected only runs the images you have checked using the platform’s own native selection UI, the asset checkboxes on Adobe Stock’s contributor portal or the row selection toggles on Shutterstock’s contributor site. This button only works on Adobe Stock and Shutterstock, because those are the only two supported platforms that expose a clean multi-select state the extension can hook into. On Zedge, Displate, TeePublic, Dreamstime, and Etsy the Process Selected button is hidden entirely and you get Process All only. If you need subset processing on one of those platforms, you run the extension multiple times against different visible slices of the upload queue.

On Adobe Stock and Shutterstock, use Process Selected when:

You want to test AI output on 2-3 images before committing credits to the full batch
Some images in the batch are already processed and you only want to fill the rest
You uploaded a mixed batch and want different Context per subset (process one subset, change Context, select the next subset, process)
You want to run enhanced quality on the tricky ones and standard on everything else

After clicking either button, the buttons swap out for a progress bar showing X / Y completed. Underneath, a Stop link halts the batch mid-run. Credits already consumed stay consumed. Nothing fills retroactively.

Three collapsed sections below the main controls. None touch the metadata pipeline. One line each:

Have a code? Promo redemption. First-time buyers use code FirstOrder for 25% off.
Refer & Earn Personal referral link. Referrals get 55 free credits, you earn recurring credits on every purchase they make (referral tiers visible in the popup).
Site not supported? On unsupported upload pages, a Scan & Request button appears. Click it to send us the page structure and opt in for an email when we ship support.

Platform-Specific Controls

Now the controls that only show up on specific platforms, or behave differently per platform. Here’s the quick matrix:

Control	Adobe Stock	Shutterstock	Zedge	Displate	TeePublic
Context field	✓	✓	✓	✓	✓
Enhanced quality	✓	✓	✓	✓	✓
Overwrite titles	always on	✓	✓	✓	✓
Allow brand names	✓	✓	✓	✓	✓
Mark as AI-generated	✓	— (hidden)	—	—	—
Process Selected	✓	✓	—	—	—
Pricing dropdown	—	—	✓	—	—

Now the detail on each platform.

Adobe Stock

Adobe Stock uses the stock-controls baseline: AI-generated toggle (yes, visible), Overwrite existing titles (no opt-out on title itself, always fills), Context, Allow brand names. Process Selected is available.

The AI fills three fields: a title capped at 70 characters (Adobe Stock truncates longer titles at the form), 15-35 keywords under Adobe’s 50-keyword maximum, and the category from Adobe’s official taxonomy. Adobe’s algorithm penalizes keyword-stuffed titles, so the generated title stays natural-language. For vectors the prompt switches to an illustration variant. For videos it handles Adobe’s separate description and keyword fields plus motion tagging.

For the deep-dive on AutoKeyWorder’s Adobe Stock behavior specifically (Process Selected workflows, AI-generated toggle use, Context strategy), see AutoKeyWorder on Adobe Stock. For the standalone Adobe Stock keyword strategy (independent of our tool), see our Adobe Stock keywords guide.

Shutterstock

Same base controls as Adobe Stock, with one removal: AI-generated toggle is hidden. Shutterstock doesn’t accept AI from contributors, so the extension doesn’t surface the option.

Shutterstock’s form has more fields than Adobe’s. Title, description paragraph, a keyword chip input (minimum 7 keywords required, maximum 50), and two category dropdowns. Shutterstock’s ranking weights descriptions more than Adobe does, so the prompt generates a distinct description, not a title-echo, an actual paragraph. Process Selected is available.

Context is especially worth using on Shutterstock because of the description. Mood and concept in your Context string change the description voice dramatically: tourism vs documentary vs editorial vs commercial all produce different language.

For AutoKeyWorder’s Shutterstock behavior specifically (description tone testing, why AI-generated is hidden, workflow), see AutoKeyWorder on Shutterstock. For Shutterstock’s ranking strategy independent of the tool, see our Shutterstock keyword guide.

Zedge (Wallpaper Pricing)

Zedge sells wallpapers and short videos and it’s the one platform with pricing controls in the popup. The Zedge-only Pricing dropdown has four values:

Skip (manual): leaves pricing untouched, you set it yourself
Free: marks the asset as free download
Video Ad: user watches an ad to unlock
Zedge Credits: user pays in Zedge’s internal credit system

Selecting Zedge Credits reveals a Price (USD) input. Minimum $0.01, step of 0.01. The extension writes the price into Zedge’s pricing form.

Pick your pricing strategy before running. Free builds download volume and follower count. Video Ad monetizes passively without gating the user. Zedge Credits at $0.99-$2.99 targets serious buyers who want the exact wallpaper.

Everything else (Context, Overwrite, Brand names) works the same as on the other platforms. But Process Selected is hidden on Zedge, you only get Process All. If you need to treat a subset differently, you handle it in multiple runs against different visible image sets.

For the deep-dive on AutoKeyWorder’s Zedge workflow (pricing tier strategy, why the mobile-first buyer changes tagging), see AutoKeyWorder on Zedge.

Displate (Metal Posters)

Base stock controls. Process Selected hidden. The prompt is retuned for Displate’s buyer: people decorating walls, not designers licensing assets. That shifts the generated language toward subject, mood, style, color palette, room aesthetic, and fandom, instead of the editorial-technical language stock platforms reward.

The extension fills Displate’s title, description, tags, and suggests collection placement. Context is useful here for themes the image alone doesn’t convey: Dark academia aesthetic, gothic library, moody changes both the title and the tag set dramatically compared to no Context at all.

Displate moderates trademark infringement aggressively, and the buyer base skews toward licensed IP references (game characters, movie scenes, band logos). Keep Allow brand/trademark names off unless you own the rights or have a license. The AI finds perfectly rankable alternatives.

For AutoKeyWorder’s Displate-specific behavior (decor-buyer prompt tuning, brand filter use, enhanced quality on illustrations), see AutoKeyWorder on Displate. For Displate’s platform-side strategy, see the full Displate metadata playbook.

TeePublic

Base stock controls. Process Selected hidden. The prompt generates titles, descriptions, and tags tuned to TeePublic’s search and the print-on-demand buyer mindset: people searching for a gift, a fandom, a hobby, or a joke. Not a stock photo.

Context is critical on TeePublic. Give the AI the design concept or niche (dog lover gift, golden retriever pun, programmer humor, bug in production, hiking enthusiast, national parks series) and the tags come back niche-specific and conversion-focused. Without Context, you get generic shirt tags that don’t rank for anything.

TeePublic caps tags at 15 per listing. The extension respects that cap and prioritizes highest-intent search terms first, because TeePublic’s search weights early tags more.

For AutoKeyWorder’s TeePublic workflow (niche Context strategy, typography-design enhanced quality), see AutoKeyWorder on TeePublic.

What I Actually Do (My Batch Workflow)

Here’s the real 40-image workflow, not the marketing version.

Upload the batch to the platform’s upload page through the platform’s native uploader
Open AutoKeyWorder, check the Detected badge matches the platform I’m on
Look at the batch: is it uniform or mixed? Uniform = one Context string, mixed = sub-batches
Fill the Context field with the specific concept for this run. Two or three nouns plus a mood. I type it fresh every batch
Check the AI-generated toggle (if on Adobe Stock and the batch is generative), otherwise leave it off
Leave Overwrite off unless I’ve typed placeholder titles that need replacing
Leave Allow brand names off. Always. Unless I’m uploading editorial to a platform that allows it
Decide on Enhanced quality: if the batch has subtle editorial work, flip it on. If it’s standard commercial shots, leave it off and save half the credits
On Adobe Stock or Shutterstock: if I want to test first, select 2 images, click Process Selected, check the output before running the rest
On Zedge/Displate/TeePublic: hit Process All, watch the progress bar, Stop if something looks wrong
Review every filled field before submitting. I edit maybe 2 out of 40. The AI is accurate but not always how I’d phrase it
Submit on the platform’s side. AutoKeyWorder never clicks the platform’s submit button

Average time for 40 images: around 4 minutes of AI processing plus review. Compare that to roughly 3.2 minutes per image typing titles, picking keywords, and selecting categories manually. On a 40-image batch, that’s 2 hours of manual work versus roughly 15 minutes including review. If you want the full math on how much manual keywording actually costs a contributor, we broke it down here.

What Sucks (Real Talk)

Every tool has trade-offs. Here’s what AutoKeyWorder doesn’t do and won’t do:

Process Selected is Adobe Stock and Shutterstock only. We’d like it on Displate and Zedge but those platforms don’t expose a multi-select state we can hook into cleanly. If you want subset processing on Zedge, you run the extension on visible subsets and reload between runs. Clunky, but it’s the platform’s UI.
AI-generated toggle is Adobe-only. Shutterstock doesn’t take AI from contributors, and the other platforms don’t have a cleanly mappable form field. If you’re uploading AI work and the toggle isn’t visible, it’s because the platform doesn’t offer the option.
It never clicks submit. By design. We fill fields, you review and submit. If you want full automation that hits publish for you, this isn’t that tool.
Credits are required. No API key bring-your-own option. Everything runs through our servers. 300 credits for $4.99 is the entry point.
AI accuracy isn’t perfect. We target “I’d edit 2 out of 40” on standard subjects. On abstract art, surreal compositions, or fine-art illustration, it’s closer to “I’d edit 5 out of 10.” Enhanced quality helps but doesn’t eliminate it.
No Etsy on the marketing page yet. Etsy is wired in the extension with its own controls, but we don’t promote it until the copy on the landing page catches up. If you sell on Etsy and stumble into it, it works.

Troubleshooting the Weird States

A few popup states trip up new users. What each one means:

Detected badge is blank or shows the wrong platform — reload the upload page. The extension resolves the platform when the page loads.
Popup is empty or shows the login screen after you’re logged in — your session expired. Sign out and back in. If it keeps happening, clear the extension’s site data in chrome://extensions.
“You’re out of credits” screen with no batch running — your credit balance hit zero. Buy credits, redeem a promo, or use the referral link shown in the panel.
Progress bar stops mid-batch and won’t continue — click Stop, then retry. Credits already consumed stay consumed. Unfilled images can be re-run.
One image fills with empty fields — rare, but if the form-fill fails after the AI call, the credit auto-refunds and you can re-run just that image using Process Selected (Adobe Stock and Shutterstock only).
Process Selected button is missing — you’re on Zedge, Displate, TeePublic, Dreamstime, or Etsy. Those platforms don’t expose selection state. Use Process All.
AI-generated toggle missing on Shutterstock — by design. Shutterstock doesn’t accept AI-generated content from contributors.

What Next

If you’re new: install the extension, claim the first-purchase discount, run a 20-image test on a platform you already sell on. If you haven’t picked a platform yet and you’re evaluating where to actually earn, this breakdown of $3,976 in AI stock photo earnings walks through the economics. Compare the output of a Context-free run against a Context-rich run on the same 5 images using Process Selected (Adobe Stock or Shutterstock). That single side-by-side is the fastest way to understand why Context matters.

If you’ve been using AutoKeyWorder but never touched Context, Overwrite, or Process Selected: try one per batch for a week. Each solves a specific problem you’re probably currently solving manually, and the combined effect on metadata quality compounds.

If you want the tool on a platform we don’t support yet: the Scan & Request This Site button on an unsupported upload page is how we prioritize. We read every request.

Every feature in that popup exists because a user asked for it. If a control is missing from your workflow, say so. That’s how the next version gets built.