Option 01
Delete creative data
Start fresh with your creative work.
Removes
- Ads and creative files
- Scripts, rounds and briefs
- Associated creative analysis
KeepsProducts, documents, accepted knowledge and architecture.
Sign in with the account for your brand.
Anglyze itself always uses English for its system, navigation and strategic structure. This choice controls the language of the detailed knowledge Anglyze builds for this brand: pains, personas, angles, interpretations, recommendations and the default brief output. Sources are never rewritten: an English ad analyzed for a Spanish account keeps its English text, and the knowledge extracted from it is recorded in Spanish.
If you bought on Whop, come in with that account. No password needed: yours is created with the email you paid with.
What each product resolves, and everything hanging from it. A gap is drawn on its own branch, dotted: there it tells you what to go find.
Two moves: take the templates, fill them in outside, and load them here when they are ready. What matters is what is inside, not what the file is called.
Dashboard · creative intelligence
See what is running, what is covered, and what to create next.
When you write a brief from here, it lands in Scripts with its code and the whole route filled in.
What the architecture shows once you cross it with what you have already shot: which angles are written and unproduced, and which are still running and deserve another round. Sorted by the same alignment checklist the builder computes, with its breakdown. The probability scores the creative’s configuration, and the best one is always proposed here, so many tie: when they do, what settles it is the material behind them — and that is what appears under the number.
Every script written in this console. One row per hook, because each hook is its own ad.
Click a status to move the hook between pending, processed and discarded.
Hand over what your brand knows, back it up, or choose what to remove.
What Anglyze knows about this brand: products, architecture, work, decisions and learnings. A document another strategist can read, with data another account can import.
Built from your records when you ask. Nothing is invented: what was never recorded says so.
The memory is a handoff, not a backup: each deletion still makes and verifies its own private copy first.
We recommend exporting a copy for your records.
Anglyze creates a private cloud backup before either deletion.
Deletion starts only after the automatic backup is complete.
Daily automatic copies are also kept for 7 days; restoring one of those is done by the administrator.
Start fresh with your creative work.
Removes
Clear this workspace, including its architecture.
Removes
You will review the scope and confirm the workspace name before deleting.
Only an administrator can deleteDeleting workspace data is reserved to an administrator of Anglyze. Extra verification (a code by e-mail or two-factor authentication) will be added.
Backup privacyRecovery backups are private to this workspace and are never used to improve Anglyze. Any use of data for learning would be a separate choice, off by default.
A micro with fewer than three pains cannot carry a round. This is that rule, made visible. Check several to delete or merge them at once: merging moves their pains to the product you choose, deleting throws them away.
Where each ad leads, with how many go to each place. The system resolves on its own the ones it can: a product page carries its identifier, and a landing page usually names the product in its address. What it cannot work out is a comparison or a collection — there are several products there, or none, and picking one would be making it up. Those are decided here once, and the decision belongs to this account.
An ad that compares two products, or shows how they are used together, is not about a new product: it is about two that already exist. That is a combo. They create themselves when a script names two products from the bank, whatever the order or the spelling: Alfa+Beta, Beta + Alfa and Alffa y Veta are the same one. You see them here, and you can create one by hand if you want it before it shows up in an ad.
Classifying is matching against what already exists. A brand that is just starting has nothing to match against, so everything falls into What is new and has to be resolved one by one — that is a queue, not a method. This goes the other way: it reads the scripts and takes from them the macros, the micros, the pains, their angles and the literal experience quotes. Nothing enters without your approval: what comes out is a proposed tree with how many creatives support each pain.
A Reddit thread, a batch of 2 and 3 star reviews, a transcript, a competitor's copy, an agency brief. It gets split, classified against the definitions and left waiting in What is new. Nothing enters the bank until you accept it.
Not connected yet. Reading the Meta ad account is not available on this console today: the ads you see come from the public libraries, through Atria.
When it is connected, read-only, it will bring per ad: spend, hook rate, hold rate and the retention curve, and which briefs went live, matched by the Task ID in the naming. That is what lets the console say which angle actually paid off.
Everything the sources bring — your ads and those of any brand being watched — that your architecture does not register yet. The console goes over what is documented every few minutes and leaves here what it finds, already categorized in your terms, with the brand it came from, without repeating what is already there and with its recommendation to approve or reject. If you approve it, it enters the bank with its code and shows up when creating scripts. If you reject it, it is kept aside. Feed the bank does not go through here: there you approve each item as you add it, and it goes straight in.
What each term means, what you have to upload for the console to be worth anything, how it is used, and which part of the job is still a person’s. All of it with an example beside it and templates to take away.
Type the brand name or paste the URL of its Meta ad library. Pick one and every ad it has comes in — live and switched off, videos, statics and carousels: what you want to filter by, you filter later, on what is already here. A brand you bring in as a competitor is registered in Competitors.
Watching is a full sweep twice a day, at 7 and at 19: it brings in what the brand has launched and marks the ones that have gone off. The first load is not waited for — it is done by hand above, which is much faster. Click a card to open its video folder, or the button to do now what the sweep would do.
What has been brought in and transcribed, before classifying anything. Classification starts here; the result shows up in Creative references.
Notion's API refuses calls made from a browser, so this console talks to it from its own server. The token never reaches this page.
One Notion row per hook, with its whole route: product, macro, micro, pain, angle, awareness, framework, naming. Sending the same script twice creates two rows, so send once per batch.
If something looks wrong, this is the order things have to be true in.
This is where the business is run, not the creative work: who pays, what plan they are on, what each account is costing and who can get into each brand. To see a brand from the inside you go in from here, and you go in to look.
Each one with its plan and what it is spending this month. Go in to look opens the brand from the inside without being able to touch it.
It is recorded as it happens, call by call: nobody can reconstruct afterwards how many tokens a script from three weeks ago used. Months from before this existed come out empty, and empty is what they are.
So you can see which one is eating the margin of the plan that pays for it.
Prices are set on the server and not here, so an old price written into the code cannot produce a wrong report that nobody notices.
A brand account reaches one brand and works inside it. An agency account reaches several and goes in to look.
Keys are stored encrypted and never come back out whole: you can tell which one is set by its last few characters, which is all you need to recognise it without being able to take it. Send it empty to remove it.
This is not a hand-written list: each point is checked against the real system every time you open this screen. A green point is standing and working in these accounts, not just programmed.
Built but not being used yet, or only working up to a point. Each one says what it is missing.
In order of what will hurt most to be without. The top is not the hardest: it is what gets missed first.
These operations are the only thing an agency account can change inside a brand. Each one says before you press it what changes and what does not, and it is recorded with your email.
The brand reads it at the top when they sign in, with your email underneath. Send it empty to remove it.
It cannot be deleted from the console: whoever can change something cannot erase the trace of having changed it.
Where each thing lives and what governs it. Not drawn: read. The structure comes out of the repository itself — the files that exist, the rules written down, the doctrines with their evidence seal — and the state is asked for live. A map nobody updates starts lying in two weeks.
The one-line description under each file is quoted from the file’s own header comment, in the language it was written in. It is not translated on purpose: a translated copy of a comment goes stale the day someone edits the original, and then the map says something the code no longer says.
What Anglyze has learned about how the market writes, read live from this workspace’s classified competitor statics every time you open this page — nothing is stored, nothing leaves the account. The same reading feeds the offer construction chips and the style block of every brief prompt. Presence and the account’s single winner bar; never a ranking, and nothing here decides for the strategist. The morning cron refreshes the library itself: review this after it runs.
Who has signed up at anglyze.com/resources: the people who left their email, took their tokens and are opening resources with them. They are not the console accounts — they live in that project, with its own rules, and everything here is written by it.
Each one with the tokens left and the resources opened. Setting a balance writes an exact figure, it does not add. Releasing a resource costs the account nothing, and taking one away gives nothing back.
The proposals published at anglyze.com: who has opened each one, what they have asked, and what we answered. They live in that project, like the portal accounts above, and everything here is written by it.
A proposal appears here as soon as it is published on the site. This screen does not create them: it follows them. Opens are counted in the client's browser, so a page served from cache still reports one.
The ones with a check close themselves — and reopen themselves if what was fixed breaks again. The rest are closed by hand. Each one says whose it is: a task with no owner is nobody's task.
Each brand is a sealed space: its own bank, its own scripts, its own Notion. No brand can see another.
A whole made-up brand — bank, its own ads and three competitors’, scripts and media metrics: spend, purchases, ROAS, CPA and retention — to show the console without touching anyone’s data. It overwrites everything in that space. When it expires, the space empties itself.
A brand account reaches one brand and works inside it. An agency reaches several and goes in read only: it sees them whole and does not touch them. Whatever it wants changed it requests from its console, naming the operation, and it is recorded.
What each plan over there means here. When someone pays, Whop sends the identifier of their plan and the console looks up in this list what applies to them: the name they will see and the monthly spending cap. Anything not on this list does not get in: this Whop also sells courses and has a free product, and they all send the same notice. It is recorded below, so if you create a plan and forget to add it here, you can see it — and as soon as you do, the nightly sweep picks up whoever had already bought.
The identifier comes from Whop and starts with plan_ — or with prod_, if you want it to cover every plan of a product.
Whop charges; what it includes is decided here. You set the credits and the accounts: the name, the price and the cycle come from Whop. The spending cap comes from the credits, so there are no two figures to reconcile. And this same list is the one the sales page reads, so a price does not live written in two places.
It does not count clicks: it looks at what has actually been done in each account. A click does not say whether something helped, and sorting by clicks puts “star an ad” above “build a round”, which is the whole product. What comes out in red at the end is what nobody has started.
The real cost of every model call, recorded as it happens. It is what to look at before pricing a plan: what a customer actually consumes, not what you think they consume. With what their plan gives them beside it, to see whether it is too much or too little.
A price or a credit allowance that changes without an explanation is an argument three months from now. This records who changed it, when and why.
The link between the payment and the account is the email, and that is where it breaks: they pay with the work one and sign up with the personal one. Their plan sits waiting. If someone has been here for days, write to them: they most likely registered with another email, and then the plan has to be moved by hand from above.
The last fifteen events, with what arrived and what was done. Without the whole body: to fix a fault you need to know which fields came, not what they said.
Five ways to start a round, and all five end in the same place: a card per creative, its hypothesis, the confirmed commission and the script. From scratch walks the architecture from the top; Gap starts from what the bank has written and nobody has tested; Iteration starts from a piece of your own that shows results; block by block starts from the video; the written command starts from your words.
Type @ to pick the product, then say what you want in your words. Anglyze reads the order against the bank —micros, pains, personas, awareness levels, pieces with a code—, opens one card per piece, recommends what the order did not say (persona only when a lens on that pain names her; focus, level, execution, references, hypothesis), writes the scripts and stops: you review and accept each one. A thesis you wrote travels as your hypothesis.
Several products can share it. If there is only one, this step is skipped.
Which of the products attacking that condition you are going to brief.
The format you want, and how many of each. Next you say where they start: from scratch, or from a lens you already have.
Output language of the scripts. By default, the account language. The language of the reference ads never decides it.
From scratch walks the whole card: who it is for, the focus, and the context that focus needs. From a lens starts from a problem you already have written for someone — it brings her, her states, her group and the route that lens is about, and asks only what the lens does not answer.
From three places: the ones you flagged while going through the library, the ones the system puts on top, or by pasting the code by hand.
One variable at a time. If you change two and the result moves, you will not know which of the two moved it.
Everything below is inherited from the original piece and is not touched: that is what makes this an iteration and not a new video. The only thing that changes is the lever you picked, marked on each one.
The territory the problem reaches.
You can pick several. Each one will carry its own split of creatives and its own pains.
Each micro with its awareness level, its creatives and its pains. One micro can come in cold and another at product level.
The only manual step from here on. The console proposes the rest and you correct it.
Output language of the scripts. By default, the account language. The language of the reference ads never decides it.
One proposal per creative, with its alignment and the reasoning. Change the angle by hand and the rest of the proposal readjusts.
Who each brief belongs to. You can assign the whole round to one agency or split it video by video.
You will be able to sign in with your email and this password, as well as with Whop.