Three moves, from SERP to sentence.
This page walks the whole machine, end to end: agents research the keyword, the findings become the brief, and the writing starts with the evidence attached. Every window below is the real product performing, and you direct all of it.
Point it at the keyword you want to win.
Type the keyword. The research agents read every page-one result the way a strategist would (every intent, every heading, every word count) and chart what it takes to outrank them. About a minute later, the evidence is on your desk.
Point it at a keyword
One field: the keyword you want to win. Pick the intent and format to target, or let the evidence pick them for you.
They read all ten ranking pages
The agents fetch page one live and read every result end to end: the same Google your buyers see, not a cached export.
The evidence gets charted
Intents weighed, formats counted, lengths measured, core topics mapped. And the gaps nobody covers, found and marked.
The evidence becomes a brief. Make it your own.
The outline assembles itself from the research: structured to beat page one, sized to the winning length, with the open gaps claimed as headings. Then it’s yours to elevate: reorder, cut, rewrite, every edit a click or a keystroke. Nothing gets drafted until you say so.
The outline assembles from the evidence
Built to beat the pages it just read: the structure page one rewards, sized to the winning length, gaps claimed as headings.
Elevate it
Drag, cut, rewrite, claim another gap. Every edit is a click or a keystroke, and the coverage checklist keeps score as you go. Take it from ranking-ready to unmistakably yours.
Ship it: two ways
Hand it to a writer as a spec-sheet brief (keyword, intent, target length, terms, outline) with a share link you can revoke anytime. Or open it in the editor and draft it yourself with the AI agents: that’s move three.
Agents draft. You direct. The research rides along.
Open the document and the research is already attached: every agent run consumes it as context, alongside your Content Rules and Writing Memory. Ask for a full draft or a surgical edit, watch it write, take over the keyboard anytime. Then the SEO panel scores the result against the same evidence.
They write with the research attached
Every run (drafting, editing, optimizing) consumes the research as context, next to your Content Rules and Writing Memory.
You direct every run
Full drafts in Agent mode, surgical edits in Targeted mode. Or just type. It’s a real editor, and the document is always yours.
Optimize and ship
The SEO panel scores the draft against the same evidence (coverage, length, terms), then export it wherever it publishes.
Three doors. One pipeline.
Research-first is the featured path, not a requirement. Start wherever today’s work starts. Every door opens into the same connected thread.
Start from a keyword
The walk you just read: agents research the SERP, you make the brief your own, the draft lands with the evidence attached.
Start from a blank page
Pick a content type and write with AI immediately. No research required. Attach research later if the piece earns it.
Start from what you have
Import a URL, Markdown, or a paste, then run the SERP against your existing page and claim everything it’s missing.
Eleven frontier models. One editor.
WordCrafter runs the current frontier. You pick the model per run, swap mid-document, and compare how they write. Metered plans or your own API keys: your call.
The pipeline is the story. This keeps it yours.
Six more pieces working the same thread: context, memory, and control that compound with every piece you ship.
Context you attach once, use everywhere
Product facts, style guides, past research: organized, tagged, attachable to any agent run.
It learns how you write
Your edits become durable memory, consolidated as you work. Every run starts closer to your voice.
House rules every run obeys
Banned words, tone, claims policy: set once, enforced on every agent run in the project.
Your best prompts, saved
Prompt-and-settings presets you reuse across documents: the setup that worked, one click away.
Every draft, kept and credited
The document’s full history, including which model and which instructions wrote each version.
Dictate, illustrate, ship
Voice input for briefs and asks, image generation in place, and clean export wherever it publishes.
Asked before the trial starts.
The practical ones: entry points, models, pricing mechanics, voice, and whose data it is.
You point it at a keyword. Research agents read every page-one result and chart the evidence: intents, formats, lengths, topics, and the gaps nobody covers. That evidence becomes a ranking-ready brief you make your own: reorder it, cut it, claim the gaps, share it. Then the writing happens with the research attached. Agents draft and edit against it, you steer every run, and the SEO panel scores the result against the same evidence.
No. Research-first is the featured path, but you can open the editor and write with AI straight away, or import existing content and optimize it. Every door leads into the same pipeline, and you can attach research to any document later.
For your target keyword, it reads the actual top-ten Google results in real time and extracts the intent mix, the content formats that rank, measured lengths and section counts, the core topics page one rewards, the content gaps nobody covers, and a structure and outline optimized to compete.
A brief is a living spec sheet synced to the research: target keyword, dominant intent, target length, a terms checklist, and the full outline, plus an editable body for anything you want to add. You can share it with a writer through a public read-only link, and revoke that link anytime.
The current frontier, from four providers: GPT-5.5 and the GPT-5.4 family from OpenAI, Claude Opus, Sonnet, and Haiku from Anthropic, Gemini Pro, Flash, and Flash-Lite from Google, and DeepSeek V4 Pro and V4 Flash. You pick the model per run, can swap mid-document, and can bring your own API keys.
Yes, import it from a URL, Markdown, or a paste, check the fidelity preview, and optionally run SERP research against it: the outline is seeded from your existing headings, so the gaps view shows exactly what your page is missing versus what ranks.
Every plan includes research credits (a SERP research run uses one) and a monthly AI usage allowance for the drafting and editing. Prefer to meter AI spend yourself? Bring your own API keys. Plans start with a 14-day free trial you can cancel anytime.
Like you. Content Rules hold your house style (banned words, tone, claims policy) and Writing Memory learns from the edits you make, so every agent run starts from how you actually write. And it’s a real editor: you can take over any sentence, anytime.
Not by nine of the eleven, and we’ll name the exception rather than bury it. WordCrafter calls the providers’ paid APIs, and OpenAI, Anthropic, and Google don’t train on data submitted through theirs. DeepSeek’s API terms are the exception, so if that matters for your content, pick any of the other nine models. Your documents and your research are yours either way: we never train on them.
Bring a keyword. Leave with the draft.
A minute in, the brief is yours to shape. The agents write against it, and you keep every sentence.