Reviewing in a chat window doesn't scale
Copy into a chatbot, describe the change, paste the answer back. Multiply by every page, deck and email in a campaign, and by every reviewer.
Content operations for marketing and creative teams
Generate, Review, Ship, Repeat
Errata is the review desk for AI-produced content. Your team marks up drafts the way an editor does, approves what's right, and every dependent page, deck, email and one-pager is updated as one governed release — with a complete record of who decided what, and why.
- Works with the AI models you already use
- Deploys to your cloud or ours
- Every decision exportable, owned by you
Plan your campaign across every webpage and surface your content touches, from inception to launch.
Create a design strategy and decide your key details once; everything in your project will adhere to your specifications.
Share and collaborate with your team and maximize their leverage. Let Errata handle the fact-checking and style-guide enforcement, and let your people generate the insight.
In one minute
One change, every document.
Built for
- Marketing agencies
- In-house creative teams
- Brand & content studios
- Product marketing
- Communications & PR
The problem
AI writes fast. Reviewing it — and keeping everything consistent — is where the time goes.
One change, twelve places
A price, a launch date, a product name, a claim legal signed off — each lives in a dozen deliverables across a dozen files. Something is always stale.
Decisions disappear
Why was that line changed? Who approved the exception? Which version of the claim is in the brochure? The answers are in someone's inbox, or nowhere.
How it works
Review like an editor. Release like a team.
Three steps, one loop — with people in charge at every one of them.
-
1
Mark it up
Draft with any AI model, or import what you already have — Word, PDF, HTML, Markdown. Highlight the words that need to change and say what's wrong, the way you'd redline a proof.
-
2
Compare and approve
The model revises the whole document against your notes and reports what it did with each one. You see the result rendered, changed passages highlighted, and flip to the previous version with a click. Nothing changes until you accept.
-
3
Propagate with confidence
Every project has a design document — how your team writes — and specifications: the approved names, prices, dates, claims and required wording. Change a fact once and Errata finds every deliverable that depends on it and proposes the matching changes as a single release for review.
Platform
Everything a content team needs to put AI to work — under control.
Highlight-to-comment review
Select text, type the note, done. The model receives the exact words you selected, and reports back on every note: addressed, or declined with a reason.
Rendered before-and-after
No diffs to decode. Changed passages appear in green on the finished page; click any one to see what it replaced.
Memory across rounds
Notes that weren't resolved carry into the next round automatically, and the model sees the full history of what you asked for and what you accepted.
Projects with a design document and specifications
Approved claims, figures and required wording live in one place per project and inform every draft the model writes for it.
Dependency awareness
Errata proposes the relationships between deliverables and your facts; your team confirms them. From then on, every fact knows where it's used.
Model choice, not model lock-in
Anthropic, OpenAI, OpenRouter — switch models per project or per team and see which one your reviewers accept more often.
Import and export
Bring in Word, PDF, HTML and Markdown. Send out PDF, Word and Markdown. Save back to the original file.
The decision record
Every prompt, note, model response, acceptance and rejection is recorded and exportable — an audit trail today, your team's private quality standard over time.
Solutions
Built for the way creative work actually gets approved.
For agencies
Run every client in its own project with its own approved facts, brand rules and model settings. Give account teams a release view — what changed, what it affected, who approved it — that stands up in a client review.
- Client-by-client isolation and export
- Reviewer roles for account, creative and compliance
- A defensible record of every approved exception
For in-house creative teams
Keep launch materials in agreement — landing page, one-pager, deck, email, FAQ, help article — as facts change through the campaign. Fewer late corrections from sales, legal and support.
- One set of specifications per launch, used by every draft
- Change a claim once; review the whole cascade
- Works alongside Notion, Google Docs and Microsoft 365
Governance & ownership
Your content. Your decisions. Your data.
Errata is designed for organizations that need to know exactly what their AI did, and to keep what they learn.
- Nothing changes without approval. Every model revision is a proposal until a person accepts it.
- Complete, exportable records. Prompts, responses, notes, outcomes and decisions, attributable to a person, a project and a model.
- Your models, your keys. Bring your own provider accounts, or use ours. Change providers without losing anything.
- Deploy where you need to. Errata cloud, or your own cloud environment for regulated work.
- No cross-customer training. Your review history is yours — isolated by default and deletable on request.
FAQ
Questions teams ask us
Do we have to move our documents into Errata?
No. Errata is where drafts are reviewed and released, not where they live. Import from the tools you use, export back to them, and save to the original file.
Which AI models can we use?
Anthropic and OpenAI directly, and any model available through OpenRouter. Choice is a setting; the review workflow and the records are the same regardless.
Is this a replacement for Notion or Google Docs?
No. Those tools manage pages. Errata manages the agreement between deliverables — what has to match, what changed, and who approved it. It sits alongside them.
Who owns the review history?
You do. It is exportable in full, isolated to your organization, never used to train models for anyone else, and deletable — including anything derived from it.
How does a team get started?
A project, a specification, and the first document — most teams are reviewing their first draft within the hour. We'll set it up with you on the demo call.
See Errata on your own launch materials.
Bring one campaign. We'll show you the review loop, the specifications, and what a governed change across every deliverable looks like.