Content just entered its agentic era.
WordCrafter’s editor writes with you, not for you. Agents draft, edit, and optimize against your research and your context, and the keyboard is yours the moment you want it. Your rules hold the line, and your feedback becomes writing memory. Every sentence keeps sounding like you.
Fixed. Saved to your memory.
If the due date passes, send a polite reminder friendly nudge the next morning.
AI made everyone a writer. Sounding like you is the hard part.
You’ve seen the article. Everyone has. It opens with “in today’s fast-paced world,” promises to delve into things, and reads exactly like the other four pieces chasing the same keyword, because the same model wrote all five. And it’s not just the same phrasing: look closer and they all give the same advice, too. The industry has a word for it now: AI slop.
In the agentic era everyone has the same models. The only input nobody else has is you. So this editor is built around exactly that. Not AI that writes content for you. AI that writes it with you: your research, your context (files, website, notes), your rules, your feedback, your keyboard.
In today’s fast-paced business world, invoicing is more crucial than ever. Let’s delve into the key steps — clear payment terms are absolutely essential. Let’s explore the basics.
In today’s fast-paced economy, getting paid is crucial for any business. Let’s delve into the world of invoicing: your payment terms must be crystal clear. Let’s explore what matters.
In today’s fast-paced digital landscape, invoices are more crucial than ever before. Let’s delve into the first rule — setting clear payment terms is essential. Let’s explore it together.
In today’s fast-paced market, professional invoicing is crucial to cash flow. Let’s delve into the details — be sure to spell out your payment terms. Let’s explore what works.
Great writers. Not editors. Not yours.
Claude and ChatGPT write genuinely well. Plenty of marketers draft in them every day. But a wall of text and a copy button is fine for average content. It’s not how you make something yours.
Creating a professional invoice is essential for getting paid on time. Start by adding your business details, an invoice number, and the date…
Next, itemize your services with clear descriptions and rates. It is also generally wise to clearly state your payment terms and the due date…
Good prose, and then a wall of text with a copy button.
Creating a professional invoice is essential for getting paid on time. Start by adding your business details, an invoice number, and the date…
Next, itemize your services with clear descriptions and rates. It is also generally wise to clearly state State your payment terms and the due date…
The same draft, but with handles on every sentence.
It starts with the research attached.
This is the same invoicing brief the research page finished building: keyword, target length, the outline you refined, the gap you claimed. One click, and it’s a draft: the outline becomes the skeleton, and the research rides along as context every agent run reads.
No research yet? Start anywhere. Pick a content type and write immediately, or import a live page by URL, Markdown, or copy & paste. The research attaches whenever you’re ready, never as a toll gate.
The outline you refined on the research side, attached and machine-readable, with 6 of 10 ranking results read in full.
Agents draft like a team doing the work.
Brief it in plain language. The agent plans the bigger jobs, reads the attached research, searches the web when it needs a fact. And when something is genuinely your call, it asks before it assumes. Every step is on the record: the plan, the sources, the reasoning.
You also choose the minds. The Thinker, the Writer, and the Image model are picked separately, from thirteen frontier models across four providers. The strategist, the wordsmith, and the illustrator don’t have to be the same brain.
Draft’s in. Checked against your rules.
Planned by Claude Fable 5.1 · Written by GPT-6 Astra
Late payments are a process problem, not a personality one. Set the expectation on the invoice itself (due date, late fee, and what happens next) so the reminder is never a surprise.
The plan shown, sources fetched, questions asked: a run you can read, not a black box.
Your draft. Take the keyboard anytime.
See a line you’d say differently? Take it. Click in mid-draft, strike the agent’s phrase, write your own. No re-prompting, no pasting into another doc to fix one sentence. Where a chat window stops, the editor begins: your hands, your ink. And the rules keep watch even when it’s you typing.
Surgical edits, offered as suggestions.
Agent mode is a conversation: you steer with what you say next. Targeted is an instrument panel: one target, one instruction, and every dial set before you press it (word count, tone, voice, even which model thinks). The change arrives as a suggestion: the old text struck, the new text inline, yours to accept or reject (one by one, or all at once).
Prefer speed? Flip the run from Suggest to Apply and the agent writes directly. The dial is yours, per run.
Late payments are a process problem, not a personality one. Set your terms on the invoice itself, and the reminder is never a surprise.
It is generally considered a good practice to make sure that your payment terms are clearly stated, because unclear State your payment terms plainly: unclear terms can often lead to delays in getting paid.
If it still isn’t paid, spell out the late fee, firmly and kindly.
One block, one instruction, one reviewable diff. The rest of the draft never moves.
Everything it knows about you, in one place.
Prompt-and-paste AI forgets you the moment the chat ends. That’s how everyone ends up sounding the same. Agent Studio is where this editor remembers: your voice, your rules, your sources, your modes. Every agent run reads it, and every entry is yours to open, edit, or delete.
A memory is one small, readable preference (told directly, learned from a writing sample, or caught in ordinary conversation), recalled by the agent into every run, so you never have to explain your voice twice.
One editor. Multiple formats.
Articles, news, commercial pages: the agent writes them all in the same editor. Long-form streams in as real blocks: paragraphs, tables, callouts, images generated in place. Commercial pages are born as typed blueprints (hero, features, social proof, call to action standing as ghosts) and fill section by section from one ask.
Tax is the line freelancers get wrong most. Charge it when your registration says you must, and show it as its own line, never folded into the fee.
| Region | On the invoice |
|---|---|
| EU (VAT) | VAT ID + rate on every line |
| US (sales tax) | state-by-state, often none for services |
| UK (VAT) | registered? 20% standard |
When in doubt, one rule holds: the invoice shows the tax, the contract explains it.
The article streams in as real blocks: paragraphs, tables, callouts, images. The page is born as a blueprint: empty sections the agent fills one by one, in the same editor, all of it yours to edit.
The whole workshop. Nothing missing.
Thirty tools working on the same draft: no add-ons, no tiers of the editor.
The editor
30+ block types
Headings to tables to columns to callouts. Shape it by hand like a modern doc editor, or let the agent do it.
Take the keyboard
Keep typing while the agent works: in Agent mode it locks only the block it’s writing and works around your edits everywhere else.
Voice input
Dictate into the draft at talking speed, or speak the edit and the agent makes it.
Suggestions
By default, the agent’s edits land as tracked changes, not silent rewrites. Review each one, or accept or reject them all at once.
Comments
Threaded comments anchored to any block, with resolution. Suggestions land in the same review list.
Version history
Every agent run saves a version, and so do Accept all and restores. Compare, pin, and restore with a preview first.
Import & optimize
Bring a published page or any draft in by URL, Markdown, or copy & paste, and the agent rewrites, expands, and sharpens it with you. Optimize it against fresh SERP research.
SEO Optimization
Scored against the research it started from: a live checklist that flips green, plus an AI-written meta description.
Export anywhere
Rich Copy & Paste that lands cleanly in Docs or WordPress, plus PDF, Markdown, and HTML. Never locked in.
The agent
Agent mode
Brief it in plain language and steer with what you say next. For bigger jobs it shows its plan before it writes.
Targeted
Point at a block, a whole section, or the whole draft, set the dials (length, tone, voice, model), and go.
Suggest or Apply
Per run: edits land as reviewable suggestions, or flip to Apply and the agent writes directly.
Asks before it assumes
When something is genuinely your call, it asks a clarifying question instead of guessing.
Draft from research
Start with the outline or page plan in place, or let the agent draft the whole piece from the research.
Content formats
Long-form articles, news, and commercial pages (product pages, service pages, landing pages), with commercial pages built from typed sections.
Recipes
32 section types for commercial pages, each with its own formula (structure, blocks, length), tuned account-wide or per project.
Image generation
Generated inline with Google’s and OpenAI’s image models, right where it goes. Pick the shape and size, or upload your own.
Web search
The agent searches the web mid-draft to check facts and pull current information, right inside the writing flow.
Content blacklist
Ban the words that sound like AI wrote it: Targeted edits and section writes steer the model away from your list.
Saved instructions
Save a prompt + settings once (model, voice, tone, length) and rerun it in one click.
Projects
Group content under shared defaults (models, edit strategy, rules) that cascade to everything in the project.
Models & behavior
Pick the Thinker, the Writer, and the Image model separately per run, each text model with its own thinking effort.
Output language
Write in the language your readers read: Auto, or English, Spanish, French, German, Italian, Portuguese, Dutch, and more.
Your own keys
Bring your own OpenAI, Anthropic, Google, or DeepSeek keys: calls on them are billed by the provider and never count toward your AI usage allowance.
Your voice
Rules
Standing guidance, always on or pulled in when it fits: voice, editorial standards, banned words. Mark one HARD and it is non-negotiable.
Writing memory
Small, readable preferences it learns from what you tell it, each one editable. A nightly tidy merges duplicates and resolves contradictions, undoably.
Learn my voice
Paste a writing sample and your style is extracted straight into memories.
Modes
Default, Brainstorm, Drafter, Reviewer: modes that reshape how the agent works. Or build your own.
Context Library
URLs, files, and notes the agent treats as ground truth, attached to any run.
Business context
Your company, product, and audience: context the agent writes with.
You, at agent speed.
Features are the flower. This is the fireball.
Your voice, at scale.
Agent drafts arrive already sounding like you (rules held, memory applied), and every correction you give sharpens the next one. Volume stops costing identity.
Always the driver.
You direct the runs, review the suggestions, and take the keyboard whenever you want it. The agents adapt around you, never the reverse.
It compounds.
Prompts forget; your editor doesn’t. The tenth draft needs fewer edits than the first, and the gap between you and prompt-and-paste keeps widening.
Bold, safely.
A version saved after every agent run and suggestions you can reject make ambition cheap. Let the agent try the big restructure: the undo is right there.
Asked before the first draft.
Straight answers: voice, control, safety, and what happens to your content.
Like you. That is the point of the architecture. Writing memory accumulates the preferences you give the agent in feedback, Learn my voice extracts your style from a pasted sample, and rules hold your standing guidance (voice, standards, banned words). Every agent run draws on all three. The more feedback you give, the more the next draft already sounds like you.
Writing memory keeps small, readable preferences (voice, vocabulary, structure, formatting, audience, facts, and things to avoid), each one visible in Agent Studio, editable, and deletable. A nightly tidy merges duplicates and resolves contradictions, and every change it makes is undoable. Delete a memory with “never capture this again” and it lands on a blocklist the agent honors.
Yes. Terminology rules map banned terms to preferred ones, and any banned term in new text is flagged as a one-click suggestion, whether the agent wrote it or you did. Mark a rule HARD and it is non-negotiable. This page’s own demo bans “penalty” in favor of “late fee”: watch the rule catch it mid-typing.
It is a real editor: more than 30 block types, drag to reorder, comments, and a cursor that is always yours. In Agent mode, type anywhere except the block it’s writing. Agent runs default to Suggest, so AI edits land as reviewable suggestions; flip to Apply when you want it to write directly. The document is a document, not a transcript.
It can’t ruin anything permanently. The editor saves a version after every agent run and before Accept all, and doc-wide undo can roll an agent run back. Version history shows every version with how much came from AI and how much from you; restore previews the exact changes first, and saves its own version before it touches anything.
Not by OpenAI, Anthropic, or Google, and we’ll name the exception rather than bury it. WordCrafter calls the providers’ paid APIs, and those three don’t train on data submitted through theirs. DeepSeek’s API terms are the exception. Picking another provider’s model covers your writing, but some background features, such as research and remembering your preferences, run on models we choose and can use DeepSeek. If that matters for your content, ask us before you rely on it. Your content and your research are yours either way: we never train on them.
No. There are three doors. Start from research and the draft begins with it attached; start from a blank page and write with the agents immediately; or import a live page by URL, Markdown, or copy & paste and optimize it. Research makes the agents sharper, but it is never required.
Thirteen frontier models across Anthropic, OpenAI, Google, and DeepSeek. And you pick the Thinker, Writer, and Image models separately, per run, each text model with its own thinking effort. Add your own keys and calls on them are billed by the provider and never count toward your AI usage allowance.
Four ways: rich Copy & Paste that lands cleanly in Google Docs or WordPress, PDF, Markdown, and HTML. Your content is never locked in.
Claude and ChatGPT are genuinely good writers. The difference is everything around the writing. A chat gives you one block of text and a copy button. WordCrafter puts the same frontier models inside a real editor: agents that edit specific blocks as reviewable suggestions, version history, comments, your research attached as context, and rules and memory that keep your voice consistent across everything you write, not one conversation.
An AI editor that writes with you, not for you. Write the piece only you could write.
The old way prompts, pastes, and publishes like everyone else. The new way writes with you, and learns.